{
  "conference": "AI Engineer World's Fair 2026",
  "model": "gemini-embedding-2-preview",
  "dimensions": 128,
  "mrl": true,
  "nativeDimensions": 3072,
  "docsUrl": "https://ai.google.dev/gemini-api/docs/models/gemini-embedding-2-preview",
  "generatedAt": "2026-06-18T20:13:47.300Z",
  "totalSpeakers": 385,
  "speakers": [
    {
      "id": "worldsfair-speaker-0",
      "conference": "worldsfair",
      "name": "Charlie Guo",
      "role": "",
      "company": "OpenAI",
      "twitter": "https://x.com/charlierguo",
      "talks": [
        {
          "title": "OpenAI Workshop with Charlie Guo",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        },
        {
          "title": "Voice Agents Can Just Do Things",
          "description": "This talk argues that speech is becoming a control plane for software rather than just audio input/output. It introduces three practical patterns—voice-to-action, systems-to-voice, and voice-to-voice—and explains where realtime reasoning and tool-calling matter, and why chained STT/LLM/TTS systems start to break down as interactions become richer.",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.046502724,
        0.021881009,
        0.009158056,
        0.0047799274,
        0.054586392,
        -0.009861052,
        -0.07858026,
        -0.025327798,
        0.031234503,
        -0.2537424,
        -0.052253895,
        -0.0792453,
        -0.020105736,
        -0.06464724,
        -0.024046874,
        0.0015460522,
        0.039406355,
        -0.1086766,
        -0.09601355,
        -0.04283498,
        -0.051750515,
        -0.09462766,
        -0.013485751,
        0.07981403,
        -0.018550478,
        0.08337121,
        0.036988292,
        0.08406863,
        -0.10376062,
        0.4051723,
        -0.072125755,
        -0.003583777,
        0.019744085,
        -0.08027043,
        0.08468447,
        -0.10834862,
        -0.03049878,
        -0.035956677,
        0.033612907,
        0.03783437,
        0.0015917299,
        -0.011814393,
        -0.07819418,
        0.04845164,
        -0.04948258,
        -0.001503686,
        -0.080717646,
        0.023558816,
        -0.031086063,
        0.0109325545,
        -0.01636616,
        0.0068381443,
        -0.04589898,
        0.017502312,
        0.047146674,
        -0.0120039815,
        -0.02464539,
        -0.0014380298,
        0.13803585,
        -0.028126827,
        -0.15417661,
        -0.039224856,
        -0.0032325513,
        0.0039983843,
        0.05565674,
        -0.0024952262,
        0.027739326,
        -0.08375402,
        -0.051907036,
        -0.079009704,
        0.036695,
        0.050181646,
        -0.016484445,
        0.01832266,
        -0.057377636,
        -0.117396064,
        0.006762825,
        -0.0006336919,
        -0.019101534,
        -0.0043770447,
        -0.025060507,
        0.12717271,
        -0.08313202,
        -0.05421491,
        0.026827633,
        -0.029386178,
        -0.056375626,
        0.06668483,
        -0.05753089,
        0.025598839,
        -0.009490761,
        -0.042727683,
        -0.0071970127,
        0.057665378,
        -0.041349616,
        0.029571174,
        -0.06724295,
        -0.013025201,
        -0.06848831,
        -0.0049725426,
        -0.019032752,
        -0.04731623,
        0.030925874,
        -0.0072023044,
        -0.03641992,
        -0.03283906,
        0.10024599,
        0.031632677,
        0.016506884,
        0.0141305365,
        -0.06754508,
        -0.63764536,
        0.009456452,
        0.002966798,
        -0.008649408,
        -0.14456747,
        0.067192405,
        0.011012776,
        0.05867578,
        0.05734401,
        0.0005765414,
        0.0453329,
        0.028246732,
        0.017335823,
        -0.0017242372,
        0.05144469,
        0.05345737,
        -0.0045958636
      ]
    },
    {
      "id": "worldsfair-speaker-1",
      "conference": "worldsfair",
      "name": "Shane Wolf",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The best SDLC is the one you build yourself: Why orchestration changes everything",
          "description": "Industry research shows AI productivity gains have plateaued at 10–15% — because today's tools only optimize the 20% of a developer's day spent writing code. The real bottlenecks are left and right of code: planning, orchestration, review, and operations.\n\nJoin Atlassian's Shane Wolf and Andrei Bocan for a hands-on deep dive into the AI-native SDLC. In this workshop, we'll move past single-player copilots and show you how Atlassian is turning Jira into an AI-native orchestration layer for the entire software development lifecycle.\n\nYou'll work with Jira AI Planner — an always-on AI technical architect that helps you rapidly vet ideas, generate estimates, and create work breakdowns — and the Jira Coding Agent, which tackles coding tasks directly from your board without ever switching to an IDE. Then we'll go further: you'll learn how to build custom automations that chain these capabilities together, transforming your Jira board into an agentic software factory where humans set intent and agents execute.\n\nYou'll walk away with a working understanding of how to move from single-player copilots to multiplayer AI workflows — where the coordination layer becomes the leverage point.\n\nWhat you'll learn:\n\nHow AI Planner turns rough ideas into well-scoped, agent-ready work breakdowns\n\nHow to trigger the Jira Coding Agent - development directly from your backlog\n\nHow to build automations that orchestrate human-agent collaboration\n\nWhy intent and context (not code generation) are the moat in an AI-native SDLC\n\nPlease bring your own laptop to this session. No pre-work is required for this workshop.",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.07228396,
        0.054675657,
        0.020521654,
        0.011270626,
        0.044080347,
        -0.003588535,
        -0.001640281,
        0.031335134,
        0.08196508,
        -0.23215221,
        -0.031214023,
        -0.05398304,
        -0.018021977,
        -0.052114245,
        -0.08376691,
        -0.017702306,
        0.09862462,
        -0.014922782,
        -0.11400055,
        -0.05188209,
        -0.07040096,
        -0.09330724,
        0.0046318322,
        0.059996445,
        -0.002164588,
        0.086783886,
        -0.0036874034,
        0.09714928,
        -0.09231555,
        0.45467186,
        -0.06462702,
        0.008075513,
        0.06926003,
        -0.05624499,
        0.04638808,
        -0.08583608,
        -0.086441144,
        -0.0075349784,
        -0.0040050205,
        0.015456673,
        -0.06356074,
        -0.030799238,
        -0.08551888,
        0.057859927,
        -0.041371413,
        -0.06863728,
        -0.04160504,
        -0.03185519,
        -0.012578869,
        -0.014709286,
        -0.07644041,
        -0.0039008027,
        0.09199348,
        -0.0020456854,
        0.038189642,
        0.021970421,
        -0.058331035,
        -0.045493994,
        0.108396016,
        -0.06738731,
        -0.079666525,
        -0.03827193,
        -0.0014364541,
        0.1079656,
        -0.03677602,
        -0.039606724,
        0.042046554,
        -0.08065684,
        0.039883416,
        -0.11181926,
        0.0072414633,
        0.08945505,
        -0.018472522,
        0.085016906,
        -0.04518647,
        -0.07013941,
        -0.031982575,
        0.057678483,
        -0.06214396,
        0.02617795,
        -0.038813714,
        0.018104644,
        -0.08446069,
        -0.019911502,
        0.03046166,
        0.019936906,
        -0.09805975,
        0.056441177,
        0.007186251,
        0.0338313,
        -0.019472478,
        0.035352588,
        0.004473556,
        0.045751125,
        -0.08610927,
        0.02908774,
        -0.017092071,
        0.065507025,
        0.013924824,
        0.043080658,
        -0.0018480563,
        -0.061432745,
        0.0011356695,
        -0.040992264,
        -0.009865806,
        -0.10901346,
        0.06268815,
        -0.048541944,
        0.0054863305,
        0.0017727294,
        0.010275296,
        -0.6017254,
        0.03083232,
        0.006769143,
        0.0025859266,
        -0.12841381,
        0.052763898,
        0.03856766,
        0.07676136,
        0.05386267,
        -0.030664314,
        0.026484957,
        -0.04320275,
        -0.025951516,
        0.03434279,
        0.04794614,
        0.02914516,
        0.031121433
      ]
    },
    {
      "id": "worldsfair-speaker-2",
      "conference": "worldsfair",
      "name": "Andrei Bocan",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The best SDLC is the one you build yourself: Why orchestration changes everything",
          "description": "Industry research shows AI productivity gains have plateaued at 10–15% — because today's tools only optimize the 20% of a developer's day spent writing code. The real bottlenecks are left and right of code: planning, orchestration, review, and operations.\n\nJoin Atlassian's Shane Wolf and Andrei Bocan for a hands-on deep dive into the AI-native SDLC. In this workshop, we'll move past single-player copilots and show you how Atlassian is turning Jira into an AI-native orchestration layer for the entire software development lifecycle.\n\nYou'll work with Jira AI Planner — an always-on AI technical architect that helps you rapidly vet ideas, generate estimates, and create work breakdowns — and the Jira Coding Agent, which tackles coding tasks directly from your board without ever switching to an IDE. Then we'll go further: you'll learn how to build custom automations that chain these capabilities together, transforming your Jira board into an agentic software factory where humans set intent and agents execute.\n\nYou'll walk away with a working understanding of how to move from single-player copilots to multiplayer AI workflows — where the coordination layer becomes the leverage point.\n\nWhat you'll learn:\n\nHow AI Planner turns rough ideas into well-scoped, agent-ready work breakdowns\n\nHow to trigger the Jira Coding Agent - development directly from your backlog\n\nHow to build automations that orchestrate human-agent collaboration\n\nWhy intent and context (not code generation) are the moat in an AI-native SDLC\n\nPlease bring your own laptop to this session. No pre-work is required for this workshop.",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.044799004,
        0.023138834,
        0.0051085725,
        -0.01226778,
        0.064697996,
        -0.008330683,
        -0.046539515,
        0.084864534,
        0.0809053,
        -0.17385902,
        -0.030931834,
        -0.012017998,
        0.04289463,
        -0.02987819,
        -0.067180954,
        0.02577095,
        0.09543943,
        -0.044938494,
        -0.074254856,
        -0.104861125,
        -0.084409036,
        -0.106841415,
        -0.008203058,
        0.03035339,
        0.027015606,
        0.08517316,
        -0.0019945905,
        0.054779097,
        -0.12242336,
        0.46462396,
        -0.08027789,
        -0.0021394326,
        0.02983674,
        -0.10210154,
        0.07717096,
        -0.08998575,
        -0.08448833,
        -0.008115864,
        0.00058335107,
        0.040977243,
        -0.085265785,
        0.016071394,
        -0.07911477,
        0.038083058,
        -0.053112842,
        -0.051768985,
        -0.059536237,
        -0.01771684,
        0.002826521,
        -0.00433782,
        -0.047895573,
        -0.019521104,
        0.062025182,
        -0.01742628,
        0.048254315,
        0.027763745,
        -0.03054365,
        -0.05038524,
        0.09306378,
        -0.050307084,
        -0.0842047,
        -0.023039684,
        0.010212479,
        0.10830581,
        -0.074640416,
        -0.029175194,
        0.05761031,
        -0.086582065,
        0.003766653,
        -0.0442117,
        0.01893766,
        0.119701155,
        0.008344691,
        0.060055874,
        -0.0154303815,
        -0.046460122,
        -0.034882057,
        0.100775264,
        -0.03847526,
        0.0039822715,
        -0.048448563,
        0.002546162,
        -0.09246999,
        -0.0134291435,
        0.016256044,
        0.05129505,
        -0.09380053,
        0.06743679,
        -0.018551487,
        0.025250424,
        -0.023333136,
        -0.021132806,
        -0.044877954,
        0.052878916,
        -0.08626918,
        0.06863222,
        -0.026069183,
        0.04819183,
        -0.00035130567,
        0.0448282,
        -0.014646201,
        -0.10487832,
        0.0020347051,
        -0.030225005,
        0.0112594385,
        -0.074289106,
        0.03684846,
        -0.09125397,
        0.01635528,
        0.012665698,
        0.06837764,
        -0.59276915,
        0.029244665,
        0.00025861076,
        0.004377395,
        -0.15071204,
        0.046246916,
        0.020023901,
        0.061503537,
        0.054724563,
        -0.0005833603,
        0.01551295,
        -0.08356716,
        -0.004892924,
        0.095385864,
        0.042883895,
        -0.002672393,
        0.018375916
      ]
    },
    {
      "id": "worldsfair-speaker-3",
      "conference": "worldsfair",
      "name": "Ignacio Martinez",
      "role": "AI Developer Advocate",
      "company": "Oracle",
      "twitter": "https://x.com/nacho_martinez",
      "talks": [
        {
          "title": "Total Recall: Agent Memory and Harness Engineering",
          "description": "In this hands-on workshop you'll build a working autonomous agent from the harness up, in a notebook, then see it live in a full working web application and leave with one that can write and run its own automations. You'll implement every surface area yourself: a set of predefined tools, persistent memory through the Oracle AI Agent Memory package, orchestration with LangChain and LangGraph, and LLM access through OCI GenAI Service, composing the full set of Oracle primitives into one harness you understand end to end.\n\nMost teams assemble that harness from a dozen disconnected services: one store for vectors, another for state, a separate reranker, a bolt-on memory layer. We take the opposite approach, on a single unified memory core. The organizing principle is optionality by default: you shouldn't have to choose your memory substrate up front. With Oracle AI Database you get file system and database memory in one place, embedding models and rerankers running inside the database kernel, and every retrieval strategy an AI workload needs without leaving the core.\n\nAnd consolidating onto one core is what keeps the whole thing tractable. You know the drill: a production harness has you holding all those moving parts in your head at once, and most of your attention goes to keeping them in sync rather than improving the agent. Pull that sprawl into a single core and the cognitive load drops. You get to think about what the agent does, not where its state lives. That's the difference between controlling your harness and renting its pieces.",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.034955677,
        0.049931865,
        0.024779953,
        0.077080905,
        0.04390233,
        -0.034549706,
        -0.09496408,
        0.019137846,
        0.051588092,
        -0.22750668,
        0.020993927,
        -0.040189303,
        -0.039212774,
        -0.042102177,
        -0.07227386,
        -0.058559928,
        0.06568114,
        -0.039928354,
        -0.076885,
        -0.08482852,
        -0.10334114,
        -0.055677354,
        -0.040183824,
        0.11000105,
        -0.04224921,
        0.10883526,
        0.017267138,
        0.07411962,
        -0.09507846,
        0.4432772,
        -0.090370305,
        0.031233914,
        0.07607964,
        -0.09938976,
        0.0045238174,
        -0.08900151,
        -0.08908772,
        -0.014207537,
        0.014733042,
        -0.004305789,
        -0.018675612,
        0.01981923,
        0.012245038,
        0.041799404,
        -0.05084721,
        -0.014184572,
        -0.058533072,
        0.026808914,
        -0.018442689,
        -0.08402649,
        -0.049453944,
        -0.010080407,
        0.06364538,
        0.051438738,
        0.05573281,
        -0.008441082,
        0.06886495,
        0.01846417,
        0.114175335,
        -0.0818481,
        -0.04992141,
        -0.06281048,
        -0.038842916,
        0.024001272,
        0.048943367,
        -0.053091463,
        0.06578402,
        -0.04607674,
        -0.035266895,
        -0.09019429,
        0.029695401,
        0.00076731015,
        -0.017396351,
        -0.017487258,
        -0.024083473,
        -0.10087508,
        -0.085516,
        0.054449134,
        -0.060817752,
        0.025390048,
        -0.024464162,
        0.023740625,
        -0.0313227,
        -0.019805891,
        0.04768045,
        0.0066074156,
        -0.056773152,
        0.06460976,
        -0.02144556,
        -0.004687458,
        0.0072899754,
        -0.005071488,
        -0.047090147,
        0.011095601,
        -0.10027972,
        -0.0042274576,
        -0.023249594,
        0.051076315,
        -0.01153728,
        0.052093588,
        -0.018574929,
        -0.038493324,
        0.04752518,
        -0.043112673,
        0.0045757284,
        -0.071570516,
        0.017247835,
        0.013624331,
        0.11096284,
        0.05571137,
        -0.03213589,
        -0.600543,
        0.016556611,
        -0.0826332,
        0.043010388,
        -0.16419344,
        0.032685578,
        -0.030401576,
        0.04478522,
        0.077213466,
        0.01091394,
        0.07668973,
        -0.007320063,
        0.07925798,
        -0.009714911,
        0.031953998,
        0.048754513,
        0.06289801
      ]
    },
    {
      "id": "worldsfair-speaker-4",
      "conference": "worldsfair",
      "name": "Du'an Lightfoot",
      "role": "Senior AI Engineer",
      "company": "Akamai Technologies",
      "twitter": "",
      "talks": [
        {
          "title": "Agents That Own Their Inference: Building Production AI Agents on Dedicated GPUs",
          "description": "Every production agent today is renting its intelligence. You're paying per token, sending your customer's data to someone else's servers, and hoping the provider doesn't rate-limit you during your launch. For most teams, that's fine. But for a growing number of teams in regulated industries, with high-volume products, latency-sensitive workloads, or rising token bills, it's starting to look like a liability.\n\nIn this 120-minute hands-on workshop you'll get a dedicated GPU and build an agent that runs on infrastructure you control. You'll stand up vLLM, point your agent at it, and drive concurrent load through the stack until you can see batching, KV cache pressure, and throughput limits in the metrics. Then you'll optimize the deployment to improve throughput while keeping per-request latency in line.\n\nThe focus isn't agent frameworks. It's the inference layer underneath them. You'll leave with working code and a real understanding of continuous batching under real concurrency, KV cache tradeoffs, vLLM's metrics, and the bottlenecks that only show up when you operate the inference server yourself.",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.036639165,
        0.06042222,
        0.014550474,
        -0.011601455,
        -0.021493195,
        0.002668337,
        -0.061466664,
        -0.029867802,
        0.069585666,
        -0.19976057,
        -0.023323337,
        -0.06401615,
        -0.041047037,
        -0.022039233,
        -0.09984838,
        0.003225154,
        0.13375674,
        -0.05518465,
        -0.05041245,
        -0.049703993,
        -0.026615053,
        -0.09261419,
        0.044251516,
        0.08022848,
        0.013242915,
        0.13408452,
        -0.04946936,
        0.12565725,
        -0.031551916,
        0.43855888,
        -0.11297903,
        -0.008986007,
        0.036572114,
        -0.07633765,
        0.055820365,
        -0.077442095,
        -0.055293404,
        -0.038149558,
        0.0321284,
        0.021085024,
        0.0032994708,
        0.057660926,
        -0.030500125,
        0.047445007,
        -0.025015019,
        0.003658917,
        -0.02466366,
        0.017715614,
        0.02053032,
        -0.0075356998,
        -0.052939367,
        -0.009792178,
        0.06029704,
        0.02849802,
        0.07061673,
        0.053353343,
        0.035773903,
        -0.08104228,
        0.130489,
        -0.04318535,
        -0.0801062,
        -0.0064677475,
        0.017778525,
        0.03992097,
        -0.004351981,
        -0.09198204,
        0.0047241044,
        -0.04326705,
        0.009529981,
        -0.049702946,
        -0.015193295,
        -0.007362074,
        -0.05009925,
        0.044187643,
        -0.056273814,
        -0.116345316,
        -0.078381345,
        0.055114392,
        0.013816163,
        0.029942296,
        -0.016820073,
        0.029053727,
        -0.033515956,
        -0.017721277,
        0.08802083,
        -0.055236,
        -0.09365587,
        -0.022254655,
        0.0019873579,
        0.019993404,
        0.0049458663,
        -0.019776838,
        -0.08859002,
        0.02312071,
        -0.06605735,
        0.037174363,
        -0.10173595,
        0.007351798,
        0.0005078432,
        -0.012771212,
        0.0023588955,
        -0.04556099,
        0.061165247,
        -0.022902766,
        0.0066609452,
        -0.042818774,
        -0.0058374875,
        0.004671774,
        0.11558426,
        -0.013582803,
        -0.016360285,
        -0.6354755,
        -0.055355005,
        -0.030174047,
        -0.039059147,
        -0.13036029,
        0.060736775,
        -0.0495794,
        0.038458887,
        0.0854349,
        0.0023466316,
        0.007363323,
        -0.029072994,
        -0.02262561,
        0.030939823,
        0.027199635,
        0.033999454,
        0.016284049
      ]
    },
    {
      "id": "worldsfair-speaker-5",
      "conference": "worldsfair",
      "name": "Zain Hasan",
      "role": "Staff AI/ML Engineer - DX",
      "company": "Together AI",
      "twitter": "",
      "talks": [
        {
          "title": "Open-Source Inference Engineering for the Agentic Era",
          "description": "Agentic coding workloads demand long contexts, multi-turn conversations, and throughput at a scale that most inference engines weren't built for. TokenSpeed is a new open-source engine purpose-built for this regime, built collaboratively by NVIDIA DevTech, AMD Triton, Qwen Inference, Together AI, and others, delivering state-of-the-art performance on NVIDIA Blackwell and AMD hardware. In this 2-hour hands-on workshop, Together Inference Research Engineers and a TokenSpeed co-creator will cover:\n\nTokenSpeed architecture and how it compares to other top open inference engines\n\nDeploying and configuring your first model\n\nOptimizing for agentic workloads: long-context, multi-turn, high-concurrency traffic\n\nKernel and hardware tuning\n\nThroughput/latency trade-offs and when TokenSpeed is the right tool",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.01713023,
        0.102446176,
        0.03181744,
        0.008073771,
        0.05135072,
        0.008516216,
        -0.062458243,
        -0.023889227,
        0.09108425,
        -0.22231099,
        -0.014654816,
        -0.054744173,
        -0.026458645,
        -0.052705586,
        -0.10698282,
        -0.013412538,
        0.056823123,
        -0.063084945,
        -0.048011124,
        -0.029194709,
        -0.03801491,
        -0.07932773,
        -0.017226942,
        0.042279992,
        -0.028490966,
        0.08182594,
        0.010866122,
        0.05008436,
        -0.0815875,
        0.42248338,
        -0.08338753,
        -0.004573671,
        0.046260145,
        -0.05605547,
        0.03470792,
        -0.10912668,
        -0.044898257,
        -0.0130110625,
        0.0011508082,
        0.013515842,
        -0.005839192,
        0.064205855,
        -0.08325505,
        0.053484906,
        -0.06416368,
        -0.067045346,
        -0.03094201,
        0.009898383,
        0.024241092,
        -0.032823153,
        -0.08142407,
        -0.009248697,
        0.09063973,
        0.0019431087,
        0.070531175,
        0.031987146,
        0.031804875,
        -0.05453901,
        0.10276332,
        -0.018378442,
        -0.14475648,
        0.01990749,
        0.008461936,
        0.012158959,
        0.016365178,
        -0.05837077,
        0.0034112986,
        -0.040941704,
        -0.08309803,
        -0.041860025,
        0.013164534,
        -0.010910788,
        -0.07170114,
        0.059527878,
        -0.05430139,
        -0.12887035,
        -0.04900941,
        0.10146594,
        0.022539524,
        0.030325353,
        -0.04598485,
        0.01530442,
        -0.0522257,
        -0.06462061,
        0.089997694,
        -0.01550444,
        -0.09517545,
        0.036660023,
        -0.02953147,
        0.04059683,
        -0.0061474317,
        -0.039252315,
        -0.0146619845,
        0.03391517,
        -0.105561696,
        0.03483824,
        -0.049640886,
        0.038311195,
        -0.012683309,
        0.0049862377,
        0.06657219,
        -0.060990352,
        0.066093616,
        -0.00010989307,
        -0.07166289,
        -0.044532992,
        0.04230991,
        -0.05776168,
        0.057330307,
        -0.019082043,
        0.022998955,
        -0.63657475,
        0.017981468,
        -0.048076592,
        -0.023718568,
        -0.14864537,
        0.02754032,
        -0.05258832,
        0.036346003,
        0.050937217,
        0.01221108,
        0.026162514,
        -0.03049814,
        0.015702967,
        -0.012315336,
        0.03303134,
        0.047569502,
        0.003015856
      ]
    },
    {
      "id": "worldsfair-speaker-6",
      "conference": "worldsfair",
      "name": "Yubo Wang",
      "role": "LLM Inference",
      "company": "Together AI",
      "twitter": "",
      "talks": [
        {
          "title": "Open-Source Inference Engineering for the Agentic Era",
          "description": "Agentic coding workloads demand long contexts, multi-turn conversations, and throughput at a scale that most inference engines weren't built for. TokenSpeed is a new open-source engine purpose-built for this regime, built collaboratively by NVIDIA DevTech, AMD Triton, Qwen Inference, Together AI, and others, delivering state-of-the-art performance on NVIDIA Blackwell and AMD hardware. In this 2-hour hands-on workshop, Together Inference Research Engineers and a TokenSpeed co-creator will cover:\n\nTokenSpeed architecture and how it compares to other top open inference engines\n\nDeploying and configuring your first model\n\nOptimizing for agentic workloads: long-context, multi-turn, high-concurrency traffic\n\nKernel and hardware tuning\n\nThroughput/latency trade-offs and when TokenSpeed is the right tool",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.057261396,
        0.077999584,
        -0.0002970145,
        0.07483931,
        0.017694784,
        -0.024483893,
        -0.10519841,
        -0.008667618,
        0.042994324,
        -0.24299376,
        -0.072499335,
        -0.0691903,
        -0.0056151254,
        0.04386597,
        -0.062089887,
        0.037769962,
        0.060951762,
        -0.061151315,
        -0.011039797,
        -0.072122686,
        -0.080905825,
        -0.063333675,
        0.010432119,
        0.060152266,
        -0.010411204,
        0.06402821,
        0.050689872,
        0.077370055,
        -0.06720728,
        0.4009873,
        -0.10139184,
        0.043842603,
        0.020418812,
        -0.0553023,
        0.06107426,
        -0.08149523,
        -0.026120415,
        -0.041558504,
        0.0039145304,
        0.040117998,
        -0.00007414144,
        0.025015974,
        -0.073261686,
        0.038051445,
        -0.08410346,
        -0.024370138,
        -0.03832044,
        -0.004191212,
        0.009669209,
        -0.009374806,
        -0.0526938,
        -0.0032785537,
        0.0633429,
        -0.006679426,
        0.05256779,
        -0.01385795,
        0.0054594884,
        -0.09428933,
        0.13953704,
        -0.012400829,
        -0.124182224,
        0.007907592,
        0.027655568,
        -0.0050754547,
        0.0028039997,
        -0.043293327,
        0.0039503877,
        -0.08960063,
        -0.070740215,
        -0.04523445,
        -0.023023173,
        -0.015155566,
        -0.06434918,
        0.054694902,
        -0.07072178,
        -0.12394543,
        -0.06430724,
        0.09449891,
        0.009827289,
        -0.031538766,
        -0.009406206,
        0.042503003,
        -0.04387726,
        -0.08175966,
        0.11673456,
        -0.0076680137,
        -0.07802163,
        0.024917992,
        -0.05712566,
        0.073777154,
        -0.010281632,
        -0.021239344,
        -0.03834268,
        0.03930718,
        -0.07342001,
        0.04139039,
        -0.04193141,
        0.044181038,
        0.0028196224,
        -0.0018351562,
        0.05026701,
        0.06281804,
        0.093781695,
        -0.039265897,
        -0.07706708,
        0.01561166,
        0.027701925,
        -0.012141035,
        0.07172109,
        0.00009461537,
        -0.023761377,
        -0.63415253,
        -0.0241213,
        -0.018321946,
        -0.047423203,
        -0.14602843,
        0.03169443,
        -0.07235383,
        0.043736648,
        0.05078899,
        -0.043174233,
        0.05560231,
        0.00084966194,
        -0.0008980181,
        -0.00658846,
        0.038899142,
        0.023787567,
        0.027122304
      ]
    },
    {
      "id": "worldsfair-speaker-7",
      "conference": "worldsfair",
      "name": "Qingyang Wu",
      "role": "Staff Research Scientist",
      "company": "Together AI",
      "twitter": "https://x.com/QingyangWu1",
      "talks": [
        {
          "title": "Open-Source Inference Engineering for the Agentic Era",
          "description": "Agentic coding workloads demand long contexts, multi-turn conversations, and throughput at a scale that most inference engines weren't built for. TokenSpeed is a new open-source engine purpose-built for this regime, built collaboratively by NVIDIA DevTech, AMD Triton, Qwen Inference, Together AI, and others, delivering state-of-the-art performance on NVIDIA Blackwell and AMD hardware. In this 2-hour hands-on workshop, Together Inference Research Engineers and a TokenSpeed co-creator will cover:\n\nTokenSpeed architecture and how it compares to other top open inference engines\n\nDeploying and configuring your first model\n\nOptimizing for agentic workloads: long-context, multi-turn, high-concurrency traffic\n\nKernel and hardware tuning\n\nThroughput/latency trade-offs and when TokenSpeed is the right tool",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.014154036,
        0.08843108,
        -0.008906509,
        0.042251855,
        0.018343313,
        -0.023044487,
        -0.056683484,
        -0.014531457,
        0.042263694,
        -0.26266998,
        -0.047934666,
        -0.035319213,
        -0.035671547,
        -0.031861063,
        -0.06817529,
        0.0168944,
        0.021530164,
        -0.05837345,
        -0.004987481,
        -0.055223368,
        -0.08028118,
        -0.08553649,
        0.013755474,
        0.042514212,
        -0.012058085,
        0.064851314,
        0.04199368,
        0.07828735,
        -0.09358036,
        0.41260567,
        -0.13723871,
        0.004633749,
        0.012094096,
        -0.040237073,
        0.097590946,
        -0.104985744,
        -0.00040030928,
        -0.059226792,
        0.026055673,
        0.024219338,
        -0.017978981,
        0.024627341,
        -0.06929165,
        0.04469482,
        -0.057100996,
        -0.037393115,
        -0.03854798,
        -0.004881245,
        -0.010455693,
        -0.014044591,
        -0.04950441,
        0.029740483,
        0.04150634,
        0.022651192,
        0.05334818,
        0.024874568,
        0.018924195,
        -0.076845944,
        0.1328939,
        -0.010073195,
        -0.10112283,
        0.016342299,
        0.03059953,
        0.018643383,
        0.024005275,
        -0.027978307,
        0.025268685,
        -0.0641544,
        -0.06389891,
        -0.043325692,
        0.02251683,
        -0.0016523657,
        -0.05098431,
        0.009923495,
        -0.09367229,
        -0.16754887,
        -0.03492337,
        0.0881291,
        0.016911173,
        -0.004280159,
        -0.04234989,
        0.03754956,
        -0.059994735,
        -0.10194741,
        0.08753951,
        0.0030824433,
        -0.11120133,
        0.028883606,
        -0.054039937,
        0.05395139,
        -0.0017740452,
        -0.059657607,
        -0.03461208,
        0.0056745247,
        -0.084954835,
        0.005634832,
        -0.052512664,
        0.0351894,
        0.008089828,
        -0.006837433,
        0.02872035,
        0.022260964,
        0.08683773,
        -0.00929523,
        -0.10188482,
        0.008385753,
        0.023100525,
        -0.015343328,
        0.07082816,
        -0.0075607984,
        -0.018821875,
        -0.6224026,
        0.030642811,
        -0.060730476,
        -0.05285457,
        -0.12980638,
        0.040240835,
        -0.08173408,
        0.07656082,
        0.06470029,
        -0.049614314,
        0.039011076,
        -0.02838458,
        0.0070415144,
        0.008385658,
        0.029124066,
        0.023361055,
        0.044445068
      ]
    },
    {
      "id": "worldsfair-speaker-8",
      "conference": "worldsfair",
      "name": "Jue Wang",
      "role": "Senior Staff Researcher",
      "company": "Together AI",
      "twitter": "",
      "talks": [
        {
          "title": "Open-Source Inference Engineering for the Agentic Era",
          "description": "Agentic coding workloads demand long contexts, multi-turn conversations, and throughput at a scale that most inference engines weren't built for. TokenSpeed is a new open-source engine purpose-built for this regime, built collaboratively by NVIDIA DevTech, AMD Triton, Qwen Inference, Together AI, and others, delivering state-of-the-art performance on NVIDIA Blackwell and AMD hardware. In this 2-hour hands-on workshop, Together Inference Research Engineers and a TokenSpeed co-creator will cover:\n\nTokenSpeed architecture and how it compares to other top open inference engines\n\nDeploying and configuring your first model\n\nOptimizing for agentic workloads: long-context, multi-turn, high-concurrency traffic\n\nKernel and hardware tuning\n\nThroughput/latency trade-offs and when TokenSpeed is the right tool",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.028613608,
        0.06597559,
        0.017427789,
        0.049479153,
        0.009490053,
        -0.011988964,
        -0.06789502,
        -0.017880367,
        0.069636926,
        -0.2531881,
        -0.081125885,
        -0.038287513,
        -0.029756542,
        0.05027749,
        -0.05857451,
        0.027775878,
        0.063778594,
        -0.08356375,
        -0.03573213,
        -0.06354652,
        -0.038892023,
        -0.025413297,
        -0.006815191,
        0.06996141,
        0.0045301383,
        0.09681007,
        0.046467155,
        0.059078515,
        -0.08883418,
        0.4149258,
        -0.109906904,
        0.009063864,
        0.02004287,
        -0.04670204,
        0.08574178,
        -0.079797536,
        -0.043020245,
        -0.037717745,
        -0.017664844,
        0.048178796,
        -0.026344005,
        0.009569045,
        -0.07788022,
        0.06293925,
        -0.08701293,
        -0.02272218,
        -0.03804925,
        0.0065777986,
        -0.005950011,
        0.0077621024,
        -0.09820488,
        0.0073302737,
        0.08573247,
        -0.00916507,
        0.027361654,
        0.019909522,
        0.010064175,
        -0.07421917,
        0.11362151,
        -0.02479747,
        -0.15243493,
        0.017842384,
        0.040993292,
        0.0024204105,
        0.010928258,
        -0.0666964,
        -0.011854334,
        -0.060934108,
        -0.0793041,
        -0.048998404,
        0.0032392326,
        -0.027671948,
        -0.04798422,
        0.043044917,
        -0.09560567,
        -0.13625418,
        -0.051607158,
        0.06814227,
        -0.0002838792,
        0.004721649,
        -0.009453375,
        0.067733504,
        -0.038940027,
        -0.052829076,
        0.08683623,
        -0.002347902,
        -0.11003214,
        0.05711639,
        -0.085568145,
        0.0540082,
        0.020094039,
        -0.02018624,
        -0.046193324,
        0.048769403,
        -0.05693793,
        0.029641021,
        -0.031323526,
        0.06670293,
        0.0115486095,
        0.028413523,
        0.047051307,
        0.053709384,
        0.06921403,
        -0.0182198,
        -0.050090615,
        -0.0038712975,
        0.019410258,
        -0.007741146,
        0.090801746,
        -0.021367261,
        -0.022412833,
        -0.6173783,
        0.028967226,
        -0.04237043,
        -0.05390023,
        -0.13021314,
        0.055354305,
        -0.07011493,
        0.030370243,
        0.08872987,
        -0.06396587,
        -0.003643,
        -0.0075216913,
        -0.015934726,
        -0.028676642,
        0.03380122,
        0.0023405405,
        0.04467561
      ]
    },
    {
      "id": "worldsfair-speaker-9",
      "conference": "worldsfair",
      "name": "Doug Guthrie",
      "role": "Solutions Engineer",
      "company": "Braintrust",
      "twitter": "",
      "talks": [
        {
          "title": "Advanced workshop: Mastering AI Observability",
          "description": "Your AI is in production, but is it actually good? In this hands-on workshop, you'll learn how to uncover patterns in your production traces using Braintrust Topics, build custom scorers to target real issues, and systematically improve your agent. By the end, you'll have a repeatable eval workflow and trace-backed evidence that your AI is actually doing what you think it is.\n\nWhat You'll Build:\n1. A quality analysis using Topics\n2. Custom scorers targeting specific issues\n3. An optimized agent with traces to prove improvements\n\nA hands-on workflow for going from \"I think my agent is working\" to trace-backed evidence that it actually is.",
          "day": "Day 1 — Workshop Day",
          "time": "9:00am-11:00am",
          "type": "session"
        }
      ],
      "embedding": [
        0.065518305,
        0.046575252,
        0.005207823,
        0.020178208,
        0.00096817716,
        -0.024524264,
        -0.10598614,
        0.0039233873,
        -0.0125375455,
        -0.20663264,
        -0.049537215,
        -0.03688275,
        0.017764192,
        -0.0027447145,
        -0.097609475,
        0.024010222,
        0.11472742,
        0.00520842,
        -0.10398883,
        -0.102236085,
        -0.08487663,
        -0.014578814,
        -0.05232325,
        0.056148153,
        0.0010692757,
        0.09852973,
        0.059869137,
        0.12837657,
        -0.01064179,
        0.45299762,
        -0.08723102,
        -0.012126906,
        0.078082986,
        0.00918422,
        0.036330614,
        -0.08547298,
        -0.034172803,
        0.0034492717,
        -0.0100906845,
        0.058316655,
        -0.08850295,
        0.06236461,
        0.0043022735,
        0.023431964,
        -0.015873056,
        -0.05842303,
        -0.04607714,
        -0.009092449,
        0.0594769,
        -0.077167496,
        0.02748669,
        -0.0043069143,
        0.047651425,
        0.033156972,
        0.056699783,
        0.04337762,
        -0.004005884,
        -0.066306725,
        0.11968718,
        0.028588863,
        -0.08886948,
        0.06275896,
        -0.007956569,
        -0.024488004,
        0.092700936,
        -0.031333398,
        0.06331039,
        -0.041771106,
        -0.10458354,
        -0.042909145,
        -0.024588238,
        -0.005993911,
        -0.077386655,
        -0.0075488854,
        -0.044525836,
        -0.120554805,
        -0.045008756,
        0.051123083,
        -0.031339824,
        -0.019230675,
        0.005361115,
        0.038644988,
        -0.07602452,
        -0.070300624,
        0.058291014,
        0.009972064,
        -0.01903611,
        0.033448305,
        -0.029777287,
        0.0514124,
        -0.0649491,
        -0.05883918,
        -0.0060160346,
        0.010335947,
        -0.057845443,
        0.030557878,
        -0.038745556,
        0.015966976,
        -0.0047914577,
        -0.0054465965,
        -0.013689506,
        -0.015186219,
        0.08140368,
        -0.01520408,
        0.039764166,
        -0.048789166,
        0.012838264,
        0.038410753,
        0.050080936,
        0.014837954,
        0.018787358,
        -0.62283564,
        0.036838867,
        -0.021176511,
        -0.031441446,
        -0.12897687,
        0.028431056,
        0.03320819,
        0.038123,
        0.03471396,
        0.00591405,
        0.06972854,
        -0.029932601,
        -0.036205452,
        -0.04385173,
        0.08747866,
        0.0926094,
        -0.00213236
      ]
    },
    {
      "id": "worldsfair-speaker-10",
      "conference": "worldsfair",
      "name": "Matt Brockman",
      "role": "AI Engineer",
      "company": "E2B",
      "twitter": "https://x.com/badphilosopher",
      "talks": [
        {
          "title": "How I learned to stop worrying and love the sandbox",
          "description": "Running sandboxes at scale can get painful. How do you manage a thousand concurrent sandboxes? We'll cover burst traffic, fast sandbox creation under load, resource exhaustion, shared state with volumes, and per-user data isolation. Then you'll trigger each failure, implement fixes, and see the cost impact in real time. You'll leave with hands-on experience debugging sandbox failures and a set of observability and scaling patterns you can start implementing.",
          "day": "Day 1 — Workshop Day",
          "time": "11:05am-12:05pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.06594109,
        0.020128753,
        0.09680387,
        -0.006066323,
        -0.045429997,
        0.030359201,
        -0.14079733,
        -0.03579258,
        0.09135212,
        -0.19077049,
        0.013569504,
        -0.043318536,
        0.046266027,
        -0.059152637,
        -0.05079181,
        0.05706055,
        0.081675775,
        -0.070970684,
        -0.07315231,
        -0.068574205,
        -0.05483817,
        -0.094722845,
        -0.02588219,
        0.0069439625,
        -0.054207914,
        0.0098560555,
        0.032105025,
        0.0431949,
        -0.08228114,
        0.46478012,
        -0.09265389,
        0.0035432854,
        0.038726725,
        -0.046169613,
        0.076651655,
        -0.10402123,
        -0.013231645,
        0.038673185,
        -0.05785365,
        0.012615598,
        0.018047282,
        0.065733284,
        -0.037741993,
        -0.008231766,
        -0.063886486,
        0.026841642,
        -0.010586076,
        -0.015091371,
        0.010684196,
        -0.025964651,
        -0.014659191,
        0.056969568,
        0.06960205,
        -0.031327862,
        0.03179155,
        0.0017250364,
        0.046648737,
        -0.092465565,
        0.1229803,
        -0.034749065,
        -0.08239501,
        0.05744556,
        -0.0020468317,
        -0.010463171,
        -0.00058529986,
        -0.08877426,
        0.021095064,
        -0.041181277,
        -0.0064707506,
        -0.04770779,
        0.047679342,
        0.066103995,
        -0.044015683,
        -0.0008293968,
        -0.08627194,
        -0.09320692,
        -0.045803636,
        0.0036963583,
        -0.035042886,
        0.009394269,
        -0.0031607999,
        0.051044375,
        -0.094573274,
        -0.006335186,
        0.08585021,
        -0.07010787,
        -0.049164966,
        0.1137209,
        0.029307947,
        0.055768985,
        -0.030280817,
        0.022848072,
        -0.048895117,
        0.05838887,
        -0.10915108,
        0.04423854,
        0.00512928,
        -0.030431474,
        0.0065382724,
        0.027367216,
        0.0050127152,
        0.045055863,
        0.048526317,
        -0.047486503,
        -0.047567617,
        -0.07904923,
        0.03217774,
        0.034427453,
        0.13149713,
        -0.061838027,
        0.048891794,
        -0.58523124,
        -0.058656123,
        0.0075364616,
        -0.041112334,
        -0.14157848,
        0.04393583,
        0.0584721,
        0.01338338,
        0.064529195,
        -0.036006883,
        0.07518925,
        -0.09357533,
        0.006027371,
        0.029195776,
        0.036281604,
        0.015153124,
        0.05966417
      ]
    },
    {
      "id": "worldsfair-speaker-11",
      "conference": "worldsfair",
      "name": "Lab B",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Microsoft - Bonus 1hr",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "11:05am-12:05pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.053890925,
        0.01974466,
        0.017773936,
        -0.04844927,
        0.008642487,
        -0.030250939,
        -0.037576713,
        -0.016035514,
        0.059558745,
        -0.2298793,
        0.017908731,
        0.05575135,
        -0.040242,
        0.065539956,
        -0.023901245,
        0.039692268,
        0.096788704,
        -0.043745674,
        -0.12914145,
        0.005067968,
        -0.09604421,
        -0.1228998,
        -0.04164308,
        0.0062135737,
        -0.016863057,
        0.076450974,
        -0.019019509,
        -0.021248858,
        -0.03986732,
        0.44656548,
        -0.05263503,
        0.004922904,
        -0.032776993,
        0.034336068,
        -0.01390682,
        -0.09336881,
        -0.019335246,
        0.027262254,
        -0.031138316,
        0.038121384,
        -0.084061585,
        0.042429756,
        -0.047669437,
        -0.03021884,
        0.016540268,
        0.029022904,
        -0.049545906,
        -0.023613965,
        -0.0066026757,
        0.07644925,
        -0.03651909,
        -0.09039303,
        0.011372573,
        0.04022792,
        0.008899463,
        0.04604425,
        -0.043227617,
        0.017483152,
        0.09811603,
        -0.010816609,
        -0.051459145,
        -0.06781848,
        -0.057791267,
        0.022261523,
        -0.07616477,
        0.067925125,
        0.123397216,
        -0.060104176,
        0.020486528,
        -0.073244385,
        0.13501135,
        0.03397391,
        -0.031068316,
        0.013810501,
        -0.061034124,
        0.00073571585,
        -0.121635966,
        0.0066943136,
        0.03059933,
        -0.006316438,
        0.028562693,
        0.07300763,
        -0.09330557,
        -0.049748883,
        0.06825078,
        0.004374916,
        -0.107806645,
        0.038550943,
        -0.028320316,
        0.0337586,
        -0.06022857,
        0.019211842,
        -0.060251016,
        0.05164756,
        -0.087758206,
        0.044359285,
        -0.030028082,
        0.05775316,
        -0.04552901,
        0.011241828,
        0.015198392,
        -0.0237179,
        -0.0028335135,
        0.024629885,
        0.025501326,
        -0.065543875,
        -0.048760567,
        0.05357059,
        0.12251126,
        0.002140346,
        0.029633252,
        -0.6081694,
        -0.022713138,
        -0.020085668,
        -0.02630935,
        -0.16327836,
        0.031975005,
        -0.034108344,
        0.05027503,
        0.066614814,
        -0.012358418,
        0.07113041,
        -0.036406666,
        -0.012571691,
        0.028896837,
        0.06446256,
        0.018321492,
        -0.007837217
      ]
    },
    {
      "id": "worldsfair-speaker-12",
      "conference": "worldsfair",
      "name": "Dhruv Nathawani",
      "role": "Senior Research Scientist",
      "company": "Nvidia",
      "twitter": "",
      "talks": [
        {
          "title": "Teaching Agents to Search: Building Synthetic Training Pipelines with NVIDIA Data Designer",
          "description": "Modern agentic systems often fail because the right training data simply does not exist. Search agents are a perfect example: if you want a model to browse the web effectively, you need high-quality multi-step trajectories that teach it how to search, refine queries, inspect sources, and recover from dead ends. In this session, attendees will learn how NVIDIA used Data Designer to build synthetic supervised fine-tuning data for search-capable Nemotron models, including how to define task structure, generate seed examples, produce realistic search trajectories, filter low-quality generations, and convert traces into training-ready records. The session will also cover BrowseComp-style tasks, tool-use rollouts, validation, dataset curation, and a reusable framework for designing custom datasets for specialized behaviors across reasoning, tool use, and domain-specific applications.",
          "day": "Day 1 — Workshop Day",
          "time": "11:05am-12:05pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.026296163,
        0.10147384,
        0.0014581485,
        0.018725274,
        0.027123632,
        -0.052749913,
        -0.07154907,
        0.0041870917,
        0.04325233,
        -0.22474585,
        -0.007022246,
        0.011034592,
        0.011745558,
        0.032391284,
        -0.09294327,
        -0.0406578,
        0.14049572,
        -0.029074678,
        -0.07609294,
        -0.110331036,
        -0.031709492,
        -0.044099238,
        -0.03887561,
        0.058396067,
        -0.018675253,
        0.09412727,
        0.012213008,
        0.0623321,
        -0.0824118,
        0.45151895,
        -0.12085596,
        -0.02220001,
        0.031181028,
        -0.03703321,
        0.07358644,
        -0.059093535,
        -0.03644792,
        -0.009715982,
        0.018002203,
        0.060726684,
        -0.025207277,
        0.04962709,
        -0.034184974,
        0.03520832,
        -0.014719058,
        -0.031032765,
        -0.032277856,
        0.022930017,
        0.0016201636,
        0.024149332,
        -0.052022442,
        0.04071846,
        0.12325839,
        -0.04065121,
        0.055591833,
        0.017155187,
        0.005776913,
        -0.08298376,
        0.045414116,
        -0.071897216,
        -0.12950371,
        0.054215312,
        0.042601906,
        0.017658776,
        0.010623665,
        -0.09084323,
        0.067720704,
        -0.0029399851,
        -0.023202531,
        -0.02458741,
        0.043240517,
        0.0097447885,
        -0.07706454,
        0.02017189,
        -0.017883733,
        -0.14818619,
        -0.055240642,
        0.11061976,
        0.007992522,
        0.022315376,
        0.030724384,
        0.030979892,
        -0.009898034,
        -0.0042210747,
        0.10534616,
        -0.0050517865,
        -0.0517379,
        0.041631438,
        -0.050798986,
        0.0471722,
        0.023701672,
        0.029227879,
        -0.063897885,
        -0.0005446991,
        -0.08658556,
        -0.018122293,
        -0.06058452,
        0.044958346,
        0.04491353,
        -0.0018592522,
        0.0781061,
        0.029552292,
        0.05512192,
        -0.011908858,
        -0.046488106,
        -0.04227374,
        0.03495911,
        0.026449133,
        0.08257613,
        -0.034774777,
        -0.014282644,
        -0.61281145,
        0.0033268125,
        -0.07043563,
        -0.055904966,
        -0.14737263,
        0.012963777,
        -0.03762183,
        0.036960736,
        0.02959014,
        -0.026583346,
        0.05456442,
        -0.0028231642,
        0.0038440211,
        -0.019138645,
        -0.042956293,
        0.02286245,
        0.025054468
      ]
    },
    {
      "id": "worldsfair-speaker-13",
      "conference": "worldsfair",
      "name": "Nnenna Ndukwe",
      "role": "Principal Developer Advocate and Software Engineer",
      "company": "Qodo AI",
      "twitter": "https://x.com/nnennahacks",
      "talks": [
        {
          "title": "How to Build Quality Gates into Agentic Coding Workflows",
          "description": "AI coding agents can now generate code at unprecedented speed. But faster code generation creates a new engineering problem: how do we know when agent-written code is actually safe, maintainable, and ready to merge? In this hands-on workshop, attendees will build an agentic coding workflow with enforceable code quality gates across planning, implementation, testing, and code review. By the end of the session, participants will have a working reference pattern for agentic software delivery: an AI-assisted workflow that can inspect a repo, implement a change, run tests, evaluate risk, respond to feedback, and surface what still requires human judgment. This is a technical enablement session for engineers building with AI coding agents, platform teams designing agentic SDLC workflows, and AI engineering leaders thinking about how to scale software quality with AI.",
          "day": "Day 1 — Workshop Day",
          "time": "11:05am-12:05pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.051022314,
        0.047305815,
        0.08871024,
        0.01986773,
        0.0049052886,
        -0.014428389,
        -0.05523364,
        -0.0033844828,
        0.055765383,
        -0.20704508,
        -0.013725144,
        0.014406393,
        0.0020086307,
        -0.001825077,
        -0.046536572,
        -0.005658672,
        0.13139533,
        -0.08118505,
        0.008010611,
        -0.05654625,
        -0.07886141,
        -0.042146996,
        0.014877056,
        0.06960297,
        -0.070464864,
        0.09911281,
        -0.023743493,
        0.034040727,
        -0.04351222,
        0.4584648,
        -0.07509094,
        0.0595596,
        0.088138334,
        -0.06471128,
        0.00889984,
        -0.06573963,
        -0.025568258,
        -0.024974618,
        0.027356518,
        0.02099912,
        -0.069075316,
        0.018916478,
        -0.07414106,
        0.03814151,
        -0.04263794,
        -0.02411286,
        -0.020688403,
        0.07963027,
        0.04031986,
        -0.02673006,
        -0.03898164,
        -0.033074785,
        0.11657787,
        0.011533397,
        0.091389544,
        0.025270272,
        0.033643827,
        -0.08611577,
        0.10037635,
        0.0038665053,
        -0.094595835,
        0.0073657176,
        -0.00000321321,
        0.03449778,
        -0.03400182,
        -0.093626104,
        0.04291051,
        -0.0649921,
        -0.037078556,
        -0.03567135,
        0.005690847,
        0.048610073,
        -0.06688738,
        0.064404495,
        -0.04169794,
        -0.15111403,
        -0.0106853265,
        0.05287064,
        -0.039217237,
        -0.00822445,
        0.000093613904,
        0.018367812,
        -0.07409689,
        -0.017051918,
        0.046599723,
        -0.020979665,
        -0.12394247,
        0.020283198,
        -0.017490396,
        -0.008271709,
        0.01140697,
        -0.01028568,
        -0.01836372,
        0.07167942,
        -0.1219732,
        0.0061429814,
        -0.03291893,
        0.008238998,
        0.020507516,
        0.038351446,
        -0.058733,
        -0.07979081,
        0.031918652,
        -0.022992248,
        0.0033530938,
        -0.007671213,
        -0.056000914,
        -0.008880703,
        0.08185735,
        0.019352743,
        0.03551003,
        -0.5972624,
        -0.009107116,
        -0.031381212,
        -0.021608895,
        -0.16085604,
        0.081117734,
        0.065059446,
        0.07354429,
        0.138683,
        -0.0041300245,
        0.074774384,
        -0.054389514,
        0.021113185,
        0.014891975,
        0.0011112985,
        0.029568426,
        -0.022866946
      ]
    },
    {
      "id": "worldsfair-speaker-14",
      "conference": "worldsfair",
      "name": "Charles Frye",
      "role": "AI Engineer",
      "company": "Modal",
      "twitter": "",
      "talks": [
        {
          "title": "What is an Inference Engine, Anyway?",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "11:05am-12:05pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        -0.021649856,
        -0.037418578,
        0.033985633,
        0.10882467,
        0.0450209,
        -0.041701563,
        -0.0378781,
        0.009282613,
        0.05640677,
        -0.21004488,
        -0.039488718,
        -0.060023587,
        -0.038638957,
        -0.04719613,
        -0.039528314,
        0.06441808,
        0.066228025,
        -0.06479294,
        -0.03108443,
        -0.07806468,
        -0.093185715,
        -0.15783444,
        -0.041178495,
        0.021611283,
        -0.059027944,
        0.078463405,
        0.0017753305,
        0.04313012,
        -0.11023789,
        0.45603475,
        -0.106731,
        -0.049921274,
        0.07053634,
        -0.03475299,
        0.043690167,
        -0.022144184,
        -0.025495663,
        0.0314854,
        -0.039462473,
        -0.006066815,
        -0.017292859,
        0.06682035,
        -0.037069894,
        -0.006124274,
        -0.07843316,
        0.0479581,
        -0.07928627,
        0.005081201,
        -0.017845573,
        0.036318783,
        -0.07238714,
        0.015495704,
        0.0235645,
        0.0040302267,
        0.052295636,
        -0.009676062,
        -0.021583423,
        -0.091855586,
        0.10011048,
        -0.09113001,
        -0.07988216,
        -0.019572433,
        -0.028335743,
        0.0110690445,
        0.10024741,
        -0.06546329,
        0.04971934,
        -0.062340204,
        -0.0033722466,
        -0.067059375,
        -0.020627996,
        -0.030943243,
        -0.03214656,
        0.05239565,
        -0.10787139,
        -0.07145552,
        -0.015250873,
        0.0013727146,
        -0.017605891,
        -0.03087977,
        0.054370664,
        0.032523964,
        -0.07546718,
        -0.09672969,
        0.036279593,
        -0.01522008,
        -0.04614679,
        0.018403502,
        0.0067996983,
        0.038697544,
        -0.012234505,
        -0.014660396,
        -0.035280608,
        0.025712391,
        -0.034902338,
        -0.047901936,
        -0.053002957,
        -0.024636528,
        0.03128408,
        0.02715893,
        -0.025080794,
        -0.017044488,
        0.08185739,
        0.018964002,
        -0.07586574,
        -0.100371614,
        0.06302737,
        0.022426799,
        0.064513095,
        -0.012432363,
        -0.0068802116,
        -0.61946523,
        -0.037098654,
        -0.03933223,
        -0.020619107,
        -0.13082188,
        0.037633937,
        -0.0076248697,
        0.006751892,
        0.06654994,
        -0.0015104008,
        0.053571433,
        -0.025091365,
        -0.0077731186,
        0.02250179,
        0.04630604,
        0.058950156,
        0.07883233
      ]
    },
    {
      "id": "worldsfair-speaker-15",
      "conference": "worldsfair",
      "name": "Tejas Kumar",
      "role": "",
      "company": "IBM",
      "twitter": "",
      "talks": [
        {
          "title": "Your Evals Are Lying to You",
          "description": "“Our evals pass and our velocity is up, so it works.” It’s the most reassuring sentence in AI engineering and also the most dangerous. Teams are shipping more code than ever while incidents per PR and change-failure rates climb, and the instruments meant to catch this are quietly broken. This talk takes apart both halves of that false comfort. First, why velocity lies: the same AI-driven throughput that lights up your dashboard is what’s eroding quality underneath it. Then we explore four ways offline evals deceive you: LLM-as-judge bias (your grader rewards confident, wordy, wrong answers over terse correct ones), staleness, distribution shift between your golden set and real traffic, and single-score evals that hide which step of an agent actually failed. The centerpiece is a live demo. We’ll wire up an LLM judge on stage and watch it crown a confident, friendly, factually wrong answer. Then we’ll fix it live on stage with a three-line rubric change. Same model, different instrument. From there we’ll build up what to measure instead: traces and spans, production observability, probe-based evaluation, error budgets, and quality leading indicators that sit beside every velocity number. Attendees will leave with a five-line checklist they can apply Monday. No prior eval tooling required. If you’ve ever shipped something agentic and had a nagging feeling the dashboards were too kind, this is for you.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.045774657,
        0.08691169,
        0.04699115,
        0.023424719,
        0.0046373312,
        -0.07318234,
        -0.06391435,
        -0.023681382,
        0.019486377,
        -0.19777402,
        -0.001991772,
        0.0033410368,
        -0.030695414,
        -0.0019485399,
        -0.08020996,
        0.049234923,
        0.06772733,
        -0.116078034,
        -0.08662853,
        -0.07698326,
        -0.0042187977,
        -0.013522756,
        -0.009027997,
        0.06427442,
        -0.033200935,
        0.080402985,
        0.026390962,
        0.1200973,
        -0.07805052,
        0.4742316,
        -0.041438527,
        -0.025603622,
        0.104709595,
        -0.06284263,
        0.006253438,
        -0.07657895,
        -0.035180997,
        0.05007776,
        0.008985557,
        0.022292195,
        -0.054808527,
        0.058648825,
        0.00093406904,
        0.04141522,
        -0.058378786,
        -0.061022967,
        -0.058955777,
        0.025045894,
        -0.029776953,
        -0.08033179,
        -0.042977538,
        -0.033304937,
        -0.007721478,
        0.0007947647,
        0.03001132,
        0.036264963,
        0.021944892,
        -0.046427418,
        0.111072,
        -0.034079928,
        -0.09947782,
        0.021390025,
        0.051945113,
        -0.003856688,
        0.0331201,
        -0.13462108,
        0.0447166,
        -0.09664562,
        -0.06887577,
        -0.025903048,
        0.0011084114,
        -0.032113563,
        -0.040075477,
        -0.029680833,
        -0.08265942,
        -0.048620902,
        -0.061614532,
        0.11278097,
        0.006149121,
        0.039209917,
        -0.013779126,
        0.05178255,
        -0.07049834,
        -0.02747956,
        0.031358518,
        0.010426075,
        -0.058026623,
        0.037383523,
        -0.0321086,
        0.090897486,
        0.035776265,
        -0.028268531,
        -0.023882603,
        0.01312666,
        -0.06840568,
        0.012523251,
        -0.07444581,
        0.019035544,
        -0.046562567,
        0.019655468,
        -0.043211907,
        -0.0061689494,
        0.077092126,
        -0.0012789534,
        0.009239387,
        -0.06359717,
        -0.014534072,
        -0.02099537,
        0.0037903723,
        0.031477585,
        0.04553255,
        -0.61191326,
        0.05906922,
        -0.012294147,
        -0.035285797,
        -0.14593811,
        0.049107116,
        0.01632825,
        0.019505983,
        0.09620996,
        -0.0025227312,
        0.054400157,
        -0.04085801,
        0.014032324,
        -0.02233747,
        0.072838515,
        0.060418457,
        -0.020452175
      ]
    },
    {
      "id": "worldsfair-speaker-16",
      "conference": "worldsfair",
      "name": "Nick Nisi",
      "role": "Developer Experience Engineer",
      "company": "WorkOS",
      "twitter": "",
      "talks": [
        {
          "title": "Lifestyles of the AI-Native: Voice-coding, agent skills, hooks and scheduled tasks",
          "description": "Most engineers are bolting AI onto a workflow that was designed for a pre-AI world. The result is a faster version of the same grind. This talk is about the other path: rebuilding the daily practice of software engineering from the ground up, around what agents are actually good at.\n\nTwo senior practitioners from WorkOS will walk through how we actually work now as AI-native engineers — not in the aspirational sense, but the literal one. We think out loud and voice-code instead of typing our way to clarity. We package recurring expertise into agent skills so we're not re-explaining context every session. We wire up hooks that fire on the events we care about, and hand off scheduled tasks to agents that run overnight, while we're away from the keyboard, or otherwise off the clock. The throughline is intentional design: deciding what a human should hold onto and what should be delegated, then building the machinery to make that real.\n\nBecause there are two of us, you'll see more than one set of habits — where our setups converge on the same patterns, and where they diverge based on how each of us thinks and works. The pitch isn't \"do more.\" It's that an AI-native setup, designed deliberately, buys back attention and protects you from the burnout that comes from treating agents as a turbocharger for an old loop. Attendees will leave with a concrete mental model for voice-driven development, a pattern for authoring reusable agent skills, and working examples of hooks and scheduled automations they can adapt the same week.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.020193247,
        0.060919713,
        0.031898323,
        -0.018975092,
        -0.0090820035,
        0.030368218,
        -0.09289972,
        -0.05788457,
        0.033951733,
        -0.21867965,
        -0.015715247,
        -0.008600806,
        -0.023532687,
        0.011900224,
        -0.07638389,
        -0.050717115,
        0.091267206,
        -0.10399353,
        -0.058316942,
        -0.075532235,
        -0.037802286,
        -0.062014215,
        -0.0017823954,
        0.02134872,
        -0.038582224,
        0.09187039,
        0.017393284,
        0.041845437,
        -0.08403506,
        0.46006966,
        -0.087541886,
        0.066845655,
        0.06050199,
        -0.034867633,
        0.048672393,
        -0.054732397,
        -0.046527233,
        0.014271276,
        -0.0023620145,
        0.055525646,
        -0.058372814,
        0.023757521,
        -0.05937318,
        -0.032048147,
        -0.019449761,
        0.003679894,
        -0.05967443,
        -0.033073407,
        0.015094005,
        0.0015889155,
        -0.046486523,
        0.011070215,
        0.08112945,
        0.058040213,
        0.07460683,
        -0.015410362,
        0.014671824,
        -0.09512481,
        0.092728615,
        0.0016471068,
        -0.103763655,
        0.037362304,
        0.025262628,
        0.072371654,
        0.004176457,
        -0.097014874,
        0.0613782,
        -0.03392071,
        0.011171091,
        -0.05645907,
        -0.001637011,
        0.08537025,
        -0.05411907,
        0.057746448,
        -0.03993684,
        -0.10102476,
        -0.021285158,
        0.024401175,
        -0.033352535,
        0.039593272,
        0.00008044647,
        0.035564266,
        -0.09847322,
        0.0043949047,
        0.028045049,
        -0.027189339,
        -0.029060839,
        0.011044384,
        0.0065380135,
        0.005682932,
        0.0151852835,
        -0.053830143,
        0.0045707445,
        0.075074516,
        -0.081846595,
        -0.02066176,
        -0.023483828,
        0.00536821,
        -0.015972903,
        -0.03336199,
        0.015199786,
        -0.04772558,
        0.053000692,
        0.004384563,
        -0.015488802,
        -0.031714708,
        0.01995761,
        -0.0785063,
        0.0753646,
        0.0070798076,
        0.054960523,
        -0.6381762,
        -0.043979574,
        -0.017941426,
        -0.036887467,
        -0.14787821,
        0.03366823,
        0.03853443,
        0.044786133,
        0.08285363,
        -0.016916553,
        0.073056415,
        -0.036112677,
        -0.005323276,
        0.035502836,
        0.09702786,
        -0.017648853,
        -0.024990179
      ]
    },
    {
      "id": "worldsfair-speaker-17",
      "conference": "worldsfair",
      "name": "Zack Proser",
      "role": "AI Engineer, Applied AI",
      "company": "WorkOS",
      "twitter": "https://x.com/zackproser",
      "talks": [
        {
          "title": "Lifestyles of the AI-Native: Voice-coding, agent skills, hooks and scheduled tasks",
          "description": "Most engineers are bolting AI onto a workflow that was designed for a pre-AI world. The result is a faster version of the same grind. This talk is about the other path: rebuilding the daily practice of software engineering from the ground up, around what agents are actually good at.\n\nTwo senior practitioners from WorkOS will walk through how we actually work now as AI-native engineers — not in the aspirational sense, but the literal one. We think out loud and voice-code instead of typing our way to clarity. We package recurring expertise into agent skills so we're not re-explaining context every session. We wire up hooks that fire on the events we care about, and hand off scheduled tasks to agents that run overnight, while we're away from the keyboard, or otherwise off the clock. The throughline is intentional design: deciding what a human should hold onto and what should be delegated, then building the machinery to make that real.\n\nBecause there are two of us, you'll see more than one set of habits — where our setups converge on the same patterns, and where they diverge based on how each of us thinks and works. The pitch isn't \"do more.\" It's that an AI-native setup, designed deliberately, buys back attention and protects you from the burnout that comes from treating agents as a turbocharger for an old loop. Attendees will leave with a concrete mental model for voice-driven development, a pattern for authoring reusable agent skills, and working examples of hooks and scheduled automations they can adapt the same week.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.06129082,
        0.030634051,
        0.07940677,
        -0.0061272536,
        0.024820268,
        0.010235594,
        -0.085100174,
        -0.021982394,
        0.030753648,
        -0.22118415,
        -0.04622244,
        -0.005041497,
        -0.024016492,
        -0.04453262,
        -0.057663027,
        -0.018895065,
        0.08519168,
        -0.048135642,
        -0.06999876,
        -0.081061915,
        -0.036780696,
        -0.11477052,
        -0.043985546,
        0.031694837,
        -0.059791088,
        0.08748144,
        0.027142083,
        0.026291052,
        -0.06217485,
        0.44460607,
        -0.11622187,
        0.051107295,
        0.022213293,
        -0.05977924,
        0.05509358,
        -0.06801798,
        -0.0639047,
        0.004835728,
        -0.015618656,
        0.06416428,
        -0.012175158,
        0.034582958,
        -0.033202846,
        0.0009947649,
        -0.0074074576,
        -0.00061179156,
        -0.029311063,
        -0.02687679,
        -0.015429399,
        -0.021986462,
        -0.045994658,
        0.0012405637,
        0.0831411,
        0.0207619,
        0.08391497,
        0.029499989,
        0.03802468,
        -0.15473253,
        0.08251974,
        -0.0075606825,
        -0.030986369,
        0.006016835,
        0.009636034,
        0.05603193,
        0.0010259558,
        -0.09668979,
        0.016994204,
        0.022805331,
        -0.008075013,
        -0.098177865,
        0.023988793,
        0.06053244,
        -0.07978293,
        0.09146985,
        -0.07202284,
        -0.10364803,
        -0.022117566,
        0.05994078,
        -0.035724834,
        0.008551728,
        -0.0071966117,
        0.03483057,
        -0.11838885,
        -0.036524657,
        0.05376726,
        -0.001997012,
        -0.03618752,
        0.018556397,
        -0.045233607,
        0.05444064,
        0.010146238,
        -0.053803243,
        -0.0383446,
        0.06395381,
        -0.09760355,
        -0.022497127,
        0.0022757568,
        0.05940042,
        0.0009923773,
        -0.012426307,
        -0.015986029,
        -0.0441428,
        0.021566728,
        -0.0006750148,
        -0.013570365,
        -0.0063791894,
        -0.00052864355,
        -0.032773662,
        0.1157648,
        0.0114603015,
        0.016077306,
        -0.6342938,
        -0.04362122,
        -0.005488129,
        -0.0028550248,
        -0.14122933,
        0.045380283,
        0.024171809,
        0.042454097,
        0.06584705,
        -0.035355907,
        0.029003965,
        -0.05967582,
        -0.021998981,
        0.060603403,
        0.06891139,
        0.019254051,
        -0.01495987
      ]
    },
    {
      "id": "worldsfair-speaker-18",
      "conference": "worldsfair",
      "name": "Harshul Jain",
      "role": "Senior Software Engineer",
      "company": "Audible Inc",
      "twitter": "",
      "talks": [
        {
          "title": "2 hr deep dive on LLM Inference at Scale — Part 1 of 2",
          "description": "Most engineers using LLMs can call an API. Far fewer can explain why their model is slow, why it's running out of memory, or how the inference engines powering every major LLM API actually work. This workshop walks through the full inference stack — from how a transformer generates a single token to serving billions of tokens a day with vLLM, SGLang, TensorRT-LLM, Ray, and KServe/llm-d. 60% explanation with live demos, 40% hands-on exercises. Attendees leave with a running vLLM server they benchmarked themselves. Based on the open-source practitioners handbook being built live at github.com/harshuljain13/llm-inference-at-scale",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        },
        {
          "title": "2 hr deep dive on LLM Inference at Scale — Part 2 of 2",
          "description": "Most engineers using LLMs can call an API. Far fewer can explain why their model is slow, why it's running out of memory, or how the inference engines powering every major LLM API actually work. This workshop walks through the full inference stack — from how a transformer generates a single token to serving billions of tokens a day with vLLM, SGLang, TensorRT-LLM, Ray, and KServe/llm-d. 60% explanation with live demos, 40% hands-on exercises. Attendees leave with a running vLLM server they benchmarked themselves. Based on the open-source practitioners handbook being built live at github.com/harshuljain13/llm-inference-at-scale",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.012217063,
        0.09156996,
        0.06561811,
        0.048965972,
        0.06432787,
        -0.041664876,
        -0.100084834,
        -0.021270214,
        0.10122894,
        -0.20607921,
        -0.018233767,
        -0.035672016,
        -0.019732086,
        -0.037746795,
        -0.07159267,
        -0.0036519256,
        0.05340715,
        -0.06689415,
        -0.068040065,
        -0.057174385,
        -0.0296476,
        -0.073115915,
        -0.036813248,
        0.07418574,
        0.015481203,
        0.0472179,
        0.024318246,
        0.015971385,
        -0.118096635,
        0.43797722,
        -0.09071242,
        -0.052221846,
        0.026511557,
        -0.077543914,
        0.09510976,
        -0.0783802,
        -0.054276787,
        -0.023182351,
        0.006662592,
        0.04968496,
        0.004718349,
        0.010554067,
        -0.0010056194,
        0.051363356,
        -0.04086108,
        -0.0046754773,
        -0.08078953,
        0.021622693,
        0.011883089,
        -0.059252895,
        -0.048658155,
        -0.03427567,
        0.030661948,
        -0.024799315,
        0.031169808,
        0.040808406,
        -0.0038258298,
        -0.08486601,
        0.122717276,
        -0.046519797,
        -0.07954704,
        -0.01997458,
        0.0148407025,
        -0.019635253,
        0.03555271,
        -0.12111087,
        0.031935047,
        -0.093747884,
        -0.04832938,
        0.05424842,
        0.016701218,
        -0.063837714,
        -0.110814705,
        0.02048147,
        -0.035229925,
        -0.1138342,
        -0.078326374,
        0.11645008,
        0.013289369,
        -0.0012247227,
        -0.015231671,
        0.059267808,
        -0.06353529,
        -0.0364741,
        0.10824585,
        -0.015274138,
        -0.07646442,
        0.026816107,
        -0.06529237,
        0.0838665,
        -0.008920791,
        -0.018138282,
        -0.053273067,
        0.017346112,
        -0.07972755,
        0.032329444,
        -0.0028579931,
        0.0054689525,
        0.0016604842,
        0.025919793,
        0.01353704,
        -0.040641166,
        0.01848071,
        0.016289197,
        -0.07679212,
        -0.04459542,
        0.004824853,
        -0.03862134,
        -0.012486268,
        0.010405172,
        -0.033783503,
        -0.63219994,
        0.019125344,
        -0.008286971,
        -0.018534085,
        -0.12367442,
        0.07196642,
        -0.03454994,
        -0.005049473,
        0.026897458,
        0.0079943985,
        0.032837473,
        -0.05006336,
        0.0021325508,
        -0.022616468,
        0.0145050585,
        0.050580062,
        0.022082588
      ]
    },
    {
      "id": "worldsfair-speaker-19",
      "conference": "worldsfair",
      "name": "Kent C. Dodds",
      "role": "",
      "company": "EpicProduct.engineer",
      "twitter": "",
      "talks": [
        {
          "title": "Build the Right Thing: Product Engineering for Software Developers",
          "description": "There is nothing quite as demoralizing as finishing a feature and realizing you built the wrong thing. The code is clean. The tests pass. The ticket is closed. And none of it matters. This is happening more often, not less. AI makes it faster and cheaper to implement, which means teams can now waste entire sprints on the wrong idea at unprecedented speed. The bottleneck is no longer \"can we build it?\" It is \"should we build it?\" and \"are we sure we understand the problem?\" This session is a condensed introduction to product engineering for builders: the skills that sit upstream and downstream of implementation. We will not try to cover everything a full-day workshop would. Instead, we will focus on the highest-leverage ideas you can apply on Monday. ### What we'll cover 1. Validate before you build Most wrong builds start with an idea that was never tested. You will learn to separate real user pain from solution-shaped requests, and practice discovery questions that surface past behavior instead of hypothetical enthusiasm. 2. Prioritize what deserves to exist Not every good idea should be built now. Especially in the AI era, \"we could build this\" is not a reason to build it. We will work through a practical prioritization lens, including the Kano model, to help you distinguish fundamentals from delighters from distractions before your team commits. 3. Own the feature, not just the PR Product engineering does not end at merge. You will leave with a clearer picture of end-to-end feature ownership: staying close to users, setting up simple feedback loops, and improving what you shipped instead of moving on to the next ticket. ### Format This is a 2–3 hour session with Kent C. Dodds. Expect focused teaching, real-world examples, and short interactive exercises and discussion. This is not a full simulation lab or a ticket-closing coding workshop. It is judgment practice for engineers who already know how to ship. ### Who this is for Software engineers (and technical builders generally) who: - Have shipped something polished that nobody wanted - Feel pressure to move fast with AI and want a better filter for what deserves to exist - Want stronger product instincts without becoming a PM - Care about owning outcomes, not just closing tasks Some software engineering experience is assumed. No particular stack is required. PMs and designers often find this valuable too. ### What you'll leave with - Discovery questions for ambiguous work - A prioritization lens you can use before committing to a build - A clearer model for feature ownership and post-ship feedback loops - Language for stakeholder conversations when requirements are unclear",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        },
        {
          "title": "Build the Right Thing: Product Engineering for Software Developers — Part 2",
          "description": "Continuation of Kent C. Dodds’s workshop: Build the Right Thing: Product Engineering for Software Developers.",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.006791779,
        0.037367724,
        0.04141359,
        -0.018644953,
        0.0065393746,
        -0.006827226,
        -0.08618182,
        -0.073281854,
        0.06483852,
        -0.16697899,
        -0.041099746,
        -0.084115654,
        0.05569376,
        -0.0686973,
        -0.06536421,
        0.059249163,
        0.07928603,
        0.0020754382,
        -0.016777137,
        -0.037404127,
        -0.08371072,
        -0.07813973,
        -0.02165515,
        0.09190089,
        0.017397404,
        0.087344304,
        -0.0213857,
        0.09324931,
        -0.08593557,
        0.46270815,
        -0.040404305,
        0.024577007,
        -0.044896938,
        0.01652336,
        0.056263644,
        0.003195031,
        -0.045572933,
        0.06431196,
        -0.017828885,
        0.07418854,
        -0.10912449,
        0.0010630795,
        -0.09784775,
        -0.010479956,
        -0.05892572,
        -0.039000385,
        -0.04248774,
        -0.031012142,
        0.010570086,
        -0.0039284723,
        -0.020885414,
        0.011855378,
        0.09813985,
        -0.0004534189,
        0.030097332,
        -0.012421616,
        -0.07815456,
        -0.073250815,
        0.106793225,
        0.04658072,
        -0.06972849,
        0.020246085,
        -0.026661899,
        -0.008623316,
        0.014598228,
        -0.05514243,
        -0.006412659,
        -0.034101423,
        -0.027406696,
        -0.0672501,
        -0.023229036,
        0.09715057,
        -0.0518868,
        0.018606305,
        -0.0692555,
        -0.0840185,
        -0.045891736,
        0.021988185,
        -0.054057945,
        0.041595515,
        0.030076325,
        -0.031977672,
        -0.06454934,
        -0.07832131,
        0.063597985,
        -0.028878177,
        -0.05560661,
        -0.00675869,
        0.019856488,
        0.088239476,
        0.018584574,
        0.011189138,
        -0.057772163,
        0.03454839,
        -0.049055453,
        0.064866245,
        0.033794705,
        0.00067235884,
        -0.040905505,
        0.03760262,
        -0.0066513564,
        -0.06645842,
        -0.0012305694,
        -0.044969007,
        -0.04126582,
        -0.032316007,
        -0.017985111,
        -0.068706065,
        0.036948428,
        0.039978415,
        0.02519079,
        -0.62900126,
        -0.067121305,
        -0.008922162,
        0.004086781,
        -0.13796066,
        0.04839659,
        -0.073283195,
        0.05453134,
        0.06117251,
        -0.050564263,
        0.01771623,
        -0.070767276,
        0.017154615,
        -0.0033492271,
        0.13757733,
        0.046479836,
        0.0052967453
      ]
    },
    {
      "id": "worldsfair-speaker-20",
      "conference": "worldsfair",
      "name": "Wolfram Ravenwolf",
      "role": "AI Evangelist",
      "company": "Weights & Biases by CoreWeave",
      "twitter": "https://x.com/WolframRvnwlf",
      "talks": [
        {
          "title": "From Zero to Leaderboard: Building an End-to-End AI Agent Evaluation Pipeline",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.004310063,
        0.07377517,
        0.07744524,
        0.0465538,
        -0.017310396,
        0.040202014,
        0.03942472,
        -0.015714407,
        0.052922007,
        -0.24996978,
        -0.085517004,
        0.011283419,
        0.0059001595,
        -0.00832771,
        -0.026078504,
        0.029178753,
        0.116213016,
        -0.022107769,
        -0.044291925,
        -0.042379413,
        -0.025695143,
        -0.025210293,
        -0.04846321,
        0.006031561,
        -0.06696414,
        0.092724495,
        -0.03850327,
        0.12123059,
        -0.052957542,
        0.51721096,
        -0.049480677,
        -0.034502372,
        0.09703496,
        -0.029915575,
        0.025432853,
        -0.07890461,
        -0.021495895,
        0.09062119,
        0.042801328,
        0.036392275,
        0.023592094,
        0.05358608,
        0.0005527679,
        -0.0036287243,
        -0.07665138,
        0.02342455,
        -0.08173437,
        0.009528566,
        0.048632797,
        -0.04103029,
        0.010266347,
        -0.06882495,
        0.13024022,
        0.025630405,
        0.07543656,
        0.016677342,
        0.05746708,
        -0.093785174,
        0.05267115,
        -0.032890454,
        -0.07721517,
        0.00030352455,
        -0.02454278,
        0.067563415,
        -0.051879603,
        -0.05208752,
        0.055510625,
        -0.038998395,
        -0.015540816,
        -0.0673025,
        0.011642474,
        0.017006474,
        -0.011740126,
        0.05874193,
        -0.005679506,
        -0.07438664,
        -0.08383102,
        0.023798674,
        -0.012435762,
        0.023651345,
        0.009367107,
        0.031004723,
        -0.050317533,
        0.014451608,
        0.0512662,
        0.0021309885,
        -0.08076623,
        0.03435478,
        -0.017761633,
        -0.036152262,
        0.05208269,
        0.02624065,
        -0.0060496083,
        -0.054171596,
        -0.038242806,
        0.016855573,
        -0.05717782,
        0.048422836,
        0.037614178,
        0.0050357627,
        -0.004876905,
        0.012309227,
        0.11908135,
        -0.08110952,
        -0.0064187613,
        -0.020129664,
        0.0049827206,
        0.031268604,
        0.015380158,
        -0.036721755,
        -0.07158212,
        -0.57771724,
        -0.033832897,
        -0.064241804,
        -0.045735497,
        -0.14869052,
        0.047609538,
        0.018980227,
        0.04494235,
        0.1169799,
        -0.00017117814,
        0.028646719,
        -0.031060502,
        0.019763231,
        0.0014852497,
        0.0023387983,
        0.02933089,
        0.03395342
      ]
    },
    {
      "id": "worldsfair-speaker-21",
      "conference": "worldsfair",
      "name": "Peter Werry",
      "role": "Founding Engineer",
      "company": "Unblocked",
      "twitter": "",
      "talks": [
        {
          "title": "Beyond RAG: Build a Relational Context Engine from Scratch",
          "description": "In this workshop we'll explore the importance of context engines in modern engineering workflows, and we'll look at why traditional RAG techniques are no longer enough to deliver the context agents need.\n\nWe'll build a structured query engine that fills the gaps left by RAG, translating natural language into validated database queries over GitHub PR and Issue data. We'll implement schema-aware prompting, identity resolution, query validation, and error-driven retry loops, and you'll walk away with a working query engine for your GitHub repository.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.08031233,
        0.041505612,
        0.08937962,
        0.005062628,
        0.021305306,
        -0.009773471,
        -0.054292638,
        -0.0014289407,
        -0.0039294125,
        -0.2432496,
        -0.04791032,
        -0.04234098,
        0.07997389,
        -0.03960529,
        -0.029113267,
        -0.019908065,
        0.058793765,
        -0.0034064718,
        -0.03929871,
        -0.07848246,
        -0.029916693,
        -0.0963935,
        -0.033326987,
        0.05982485,
        -0.053136785,
        0.042680662,
        -0.07004872,
        0.102382466,
        -0.06385492,
        0.48967105,
        -0.058306485,
        -0.012922819,
        0.04612422,
        -0.022596695,
        0.05618037,
        -0.099774465,
        -0.04371173,
        0.03055367,
        -0.018082917,
        0.025665592,
        -0.024488917,
        0.0011402236,
        -0.010770913,
        0.0030096155,
        -0.054806218,
        0.020969082,
        -0.03607983,
        0.029567711,
        0.04913054,
        -0.020330029,
        -0.11839955,
        -0.0045454768,
        0.119102746,
        0.0073894127,
        0.088037096,
        0.0062046675,
        -0.024129968,
        -0.019875072,
        0.045876797,
        -0.004106415,
        -0.03877719,
        0.020835215,
        0.004580896,
        -0.029175017,
        -0.054529767,
        -0.040529057,
        0.061652794,
        0.0090216175,
        -0.0027191683,
        -0.03765794,
        0.03325826,
        -0.03827925,
        -0.052223712,
        0.039190233,
        -0.052808426,
        -0.04847802,
        -0.082662284,
        0.013408635,
        -0.04737685,
        0.036681555,
        0.019480854,
        0.03778988,
        -0.075177684,
        -0.071669586,
        0.05620657,
        -0.009294636,
        -0.066058904,
        0.023426887,
        -0.0021905063,
        0.04825229,
        0.00033821832,
        -0.012329613,
        -0.11061154,
        0.00042326894,
        -0.089139916,
        0.007054919,
        -0.037742853,
        0.0022761486,
        -0.016725766,
        -0.016703868,
        -0.028177911,
        0.019604383,
        0.019171147,
        -0.08213991,
        -0.0040997867,
        -0.033462763,
        -0.015724959,
        -0.028502796,
        0.04368029,
        -0.0003714039,
        -0.035411164,
        -0.63856643,
        -0.009195625,
        -0.028073877,
        0.015127992,
        -0.124769464,
        0.08536241,
        -0.026313517,
        0.015090593,
        0.07537926,
        -0.014079006,
        0.07787778,
        -0.039773628,
        -0.033736043,
        -0.030741934,
        0.05768374,
        -0.01230624,
        -0.015994621
      ]
    },
    {
      "id": "worldsfair-speaker-22",
      "conference": "worldsfair",
      "name": "Elizabeth Fuentes Leone",
      "role": "Developer Advocate",
      "company": "Amazon Web Services",
      "twitter": "https://x.com/lizfue",
      "talks": [
        {
          "title": "Agent Speedrun: Idea → Code → Deploy → Observe, Fix → Ship",
          "description": "One agent. Fully deployed to production before the workshop ends. We'll take you from a blank file to a running production agent using Amazon Bedrock AgentCore and Strands Agents, covering the full lifecycle: ideation, coding the agent loop, deploying to serverless infrastructure, wiring up observability, breaking it intentionally, fixing it with tracing data, and shipping the final version.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.052970376,
        0.045988463,
        0.04553188,
        -0.00032416548,
        0.01408869,
        0.0051140096,
        -0.068657085,
        -0.03092916,
        -0.006287211,
        -0.17454086,
        -0.009978852,
        -0.06374514,
        0.03726624,
        0.019588238,
        -0.051714253,
        0.019901372,
        0.1291954,
        -0.080373995,
        -0.034993507,
        -0.04454957,
        -0.08313142,
        -0.09618352,
        -0.0395251,
        0.055297535,
        -0.049393453,
        0.12855989,
        -0.05964052,
        0.058311813,
        -0.09967199,
        0.4712092,
        -0.072412595,
        -0.009967928,
        0.081534,
        -0.030611446,
        0.024809131,
        -0.08164684,
        -0.03714868,
        -0.013510287,
        0.0344714,
        -0.034364052,
        -0.031141773,
        0.06716891,
        -0.014071844,
        0.0113333035,
        -0.074724786,
        -0.0073813107,
        -0.07352886,
        0.030535249,
        0.007577673,
        0.053268317,
        -0.027814813,
        -0.017094785,
        0.07571644,
        0.047223166,
        0.053305384,
        0.027546655,
        0.029782886,
        -0.07802921,
        0.072256416,
        -0.0041879388,
        -0.07110632,
        -0.023758154,
        0.0045840233,
        0.04777281,
        -0.00087531825,
        -0.09951208,
        0.043971315,
        -0.053254187,
        -0.019639848,
        -0.09598479,
        0.030330315,
        0.026729034,
        0.0010218747,
        -0.036628805,
        -0.040201068,
        -0.14100164,
        -0.043202095,
        0.090304516,
        -0.042245872,
        0.02743504,
        -0.01780423,
        0.05142204,
        -0.13540278,
        -0.013170077,
        0.02052565,
        0.03991796,
        -0.05225593,
        0.08113046,
        -0.053173926,
        0.016212268,
        -0.004187311,
        0.04006328,
        -0.040107857,
        0.030823555,
        -0.10792454,
        0.046044726,
        -0.0015017613,
        -0.018275313,
        0.032759354,
        0.008298581,
        -0.0255867,
        0.027383352,
        0.002049372,
        -0.054112542,
        0.008405213,
        -0.058350507,
        -0.024247563,
        0.053698182,
        0.12200475,
        -0.019045241,
        -0.0010886105,
        -0.60731757,
        -0.026619254,
        -0.10574419,
        -0.009920888,
        -0.13611104,
        -0.016345339,
        -0.017365113,
        0.036416106,
        0.063735664,
        -0.024172321,
        0.10244007,
        -0.048938703,
        0.019376995,
        0.042574022,
        0.0117810825,
        -0.0193673,
        -0.018403376
      ]
    },
    {
      "id": "worldsfair-speaker-23",
      "conference": "worldsfair",
      "name": "Sandhya Subramani",
      "role": "Senior Developer Advocate for Generative AI",
      "company": "Amazon Web Services",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Speedrun: Idea → Code → Deploy → Observe, Fix → Ship",
          "description": "One agent. Fully deployed to production before the workshop ends. We'll take you from a blank file to a running production agent using Amazon Bedrock AgentCore and Strands Agents, covering the full lifecycle: ideation, coding the agent loop, deploying to serverless infrastructure, wiring up observability, breaking it intentionally, fixing it with tracing data, and shipping the final version.",
          "day": "Day 1 — Workshop Day",
          "time": "12:10pm-1:10pm",
          "type": "workshop"
        },
        {
          "title": "Tell the Robot What You Want",
          "description": "What if you could command a robot just by talking to it? This session introduces an open-source agentic AI framework that lets developers control physical sensors and actuators using natural language, by exposing hardware as programmable agent tools through a unified interface. The agent interprets the request, selects appropriate tools, and orchestrates execution. We explore a hybrid model where low-latency perception and actuation run locally on edge hardware, and higher-level reasoning and multi-step planning are delegated to cloud-based agents when needed. This preserves real-time responsiveness while enabling richer reasoning. A live robot demonstration anchors the session. Using the SO101 robotic arm powered by NVIDIA GR00T on Jetson hardware alongside HuggingFace LeRobot, attendees see how an instruction such as \"place the apple in the basket\" moves from conversation to perception to physical action.",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "sponsor"
        },
        {
          "title": "Agents That Forge Their Own Tools: Self-Modifying AI in the Wild",
          "description": "What happens when your agent decides its existing tools aren't good enough and writes new ones? Self-modifying agents can generate, test, and deploy their own tool implementations at runtime, adapting to problems they weren't explicitly programmed to solve. In this session, we'll demo a live agent that forges its own tools on the fly, discuss the safety boundaries you need, and explore where this pattern makes sense (and where it absolutely doesn't).",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.012009718,
        0.03750937,
        0.032910384,
        0.010017998,
        0.0569489,
        -0.0002227788,
        -0.09113497,
        -0.015825111,
        0.04706069,
        -0.19825517,
        -0.008998318,
        -0.04172316,
        0.019400692,
        -0.05820332,
        -0.06815436,
        -0.000014165211,
        0.096631914,
        -0.13021961,
        -0.023004256,
        -0.096393116,
        -0.031968206,
        -0.102024384,
        -0.03504543,
        0.03588176,
        -0.04415194,
        0.087349646,
        -0.01484993,
        0.0600118,
        -0.087432206,
        0.46099377,
        -0.095498554,
        -0.015994865,
        0.08091448,
        -0.05273161,
        0.06516797,
        -0.07538871,
        -0.020873552,
        -0.03958103,
        0.0044625322,
        -0.031650774,
        -0.040807817,
        0.04016851,
        -0.041284055,
        0.042294394,
        -0.012580611,
        -0.009485122,
        -0.06548138,
        0.044845276,
        -0.028058829,
        -0.017112628,
        -0.052087925,
        -0.044085454,
        0.08792681,
        0.017645994,
        0.021425962,
        0.021084568,
        0.059290137,
        -0.10152572,
        0.1233462,
        -0.013758956,
        -0.08934348,
        -0.051747218,
        0.013685389,
        0.030406123,
        -0.01440803,
        -0.084933735,
        0.042735692,
        -0.028203629,
        -0.015752528,
        -0.05748651,
        0.04059062,
        0.027259316,
        -0.0356898,
        0.026880855,
        -0.024008568,
        -0.17138308,
        -0.09258138,
        0.06781778,
        -0.018107327,
        0.038895845,
        -0.040576123,
        0.08568964,
        -0.09707984,
        -0.029962134,
        0.04161969,
        0.015722081,
        -0.06588149,
        0.046443384,
        -0.044688657,
        0.013418732,
        0.070011936,
        0.013113468,
        -0.055102248,
        0.0306584,
        -0.08624456,
        0.032703638,
        -0.037943747,
        0.0057908865,
        0.0035252844,
        0.041831307,
        -0.011903888,
        -0.022933116,
        -0.019455418,
        -0.041815016,
        -0.037264787,
        -0.03542062,
        0.04648122,
        -0.008353784,
        0.049667347,
        -0.042156875,
        0.036923196,
        -0.614764,
        -0.028917884,
        -0.013872722,
        -0.026486125,
        -0.09207412,
        0.050009675,
        -0.065662146,
        0.096483015,
        0.066375494,
        -0.014849755,
        0.09254239,
        -0.02840494,
        -0.031651553,
        -0.047090404,
        -0.0248612,
        0.0026582435,
        -0.018747278
      ]
    },
    {
      "id": "worldsfair-speaker-24",
      "conference": "worldsfair",
      "name": "Pamela Fox",
      "role": "Principal Cloud Advocate",
      "company": "Microsoft",
      "twitter": "https://x.com/pamelafox",
      "talks": [
        {
          "title": "The model swap workshop",
          "description": "Frontier labs are releasing new models constantly, and it is hard to know when “better” is better enough to justify touching a working system. On top of that, “just swap the model” often turns into real work because providers expose different APIs and different expectations around tools and structured outputs. The model swap workshop is a hands-on bake-off across frontier LLMs. We will run the same scenarios using multiple models (OpenAI, Anthropic, Kimi, and more) and compare results side by side for agentic tool use, structured outputs, and multimodal tasks. Swapping models is not just changing a model name. In this workshop, you will actually do the swaps, including moving between OpenAI-style Responses APIs and Anthropic-style Messages APIs, then see what breaks and what needs to change in your prompts, tool definitions, and JSON strategies. We will finish by running a small eval suite so you can quantify tradeoffs instead of relying on vibes. We will provide the Microsoft Foundry environment for access to the models, no account needed.",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.06988929,
        0.04944107,
        0.07406097,
        0.018661803,
        0.0042969785,
        0.07907798,
        -0.05999851,
        -0.028854912,
        0.09559986,
        -0.16637832,
        0.03571661,
        -0.07742707,
        0.014461109,
        -0.031881105,
        -0.09631503,
        0.07082794,
        0.020492304,
        -0.027295338,
        -0.09690886,
        -0.005895998,
        -0.07834569,
        -0.0835132,
        0.016819686,
        0.06930675,
        -0.021050388,
        0.067197315,
        0.005340121,
        0.01508979,
        -0.09477639,
        0.4625086,
        -0.12649524,
        0.04033162,
        0.03096042,
        -0.046413682,
        0.074893914,
        -0.090928815,
        -0.022705907,
        0.024412159,
        0.0154719055,
        -0.051767047,
        0.014310097,
        0.0026762865,
        -0.09352789,
        0.020578224,
        -0.08491985,
        -0.019827208,
        -0.059042238,
        0.011749485,
        0.039107822,
        0.075754195,
        -0.021664329,
        -0.062374204,
        0.043356836,
        -0.018883837,
        0.071096726,
        0.032512706,
        -0.0046508047,
        0.0019413966,
        0.10883638,
        0.021739714,
        -0.06799451,
        0.0016136173,
        -0.067704484,
        0.012903776,
        0.045166466,
        -0.033800386,
        0.06289921,
        -0.08696046,
        -0.03588551,
        -0.05239649,
        0.08342698,
        0.048711035,
        -0.0023007507,
        0.05805234,
        -0.061423954,
        -0.09973078,
        -0.057815053,
        0.042608432,
        0.019086627,
        0.0116112055,
        0.06387129,
        0.059716344,
        -0.1203283,
        -0.064814806,
        0.058996286,
        -0.0029469796,
        -0.10728741,
        0.06732249,
        -0.05101495,
        0.03846051,
        -0.01938088,
        -0.005973764,
        -0.09254694,
        0.022796258,
        -0.06855277,
        0.051359892,
        -0.017522538,
        0.09151056,
        -0.02512548,
        0.024911555,
        0.011722275,
        -0.051638693,
        0.0057076006,
        -0.0631391,
        -0.04786894,
        -0.070231,
        0.010914755,
        0.05907904,
        0.07838417,
        0.004478068,
        -0.03942796,
        -0.5791041,
        0.009406102,
        0.007403008,
        0.020363651,
        -0.13832097,
        0.12142371,
        -0.031364933,
        0.044638366,
        0.1359839,
        0.0032184764,
        0.037374616,
        -0.019960675,
        0.0032239396,
        -0.020719271,
        0.048437685,
        -0.024256691,
        0.09655537
      ]
    },
    {
      "id": "worldsfair-speaker-25",
      "conference": "worldsfair",
      "name": "Arun Sekhar",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The model swap workshop",
          "description": "Frontier labs are releasing new models constantly, and it is hard to know when “better” is better enough to justify touching a working system. On top of that, “just swap the model” often turns into real work because providers expose different APIs and different expectations around tools and structured outputs. The model swap workshop is a hands-on bake-off across frontier LLMs. We will run the same scenarios using multiple models (OpenAI, Anthropic, Kimi, and more) and compare results side by side for agentic tool use, structured outputs, and multimodal tasks. Swapping models is not just changing a model name. In this workshop, you will actually do the swaps, including moving between OpenAI-style Responses APIs and Anthropic-style Messages APIs, then see what breaks and what needs to change in your prompts, tool definitions, and JSON strategies. We will finish by running a small eval suite so you can quantify tradeoffs instead of relying on vibes. We will provide the Microsoft Foundry environment for access to the models, no account needed.",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.05607403,
        0.07305351,
        0.05527813,
        0.03159285,
        0.059020746,
        -0.0073793754,
        -0.07141837,
        -0.005628614,
        0.08679258,
        -0.19918925,
        0.028602844,
        -0.08310976,
        -0.013558137,
        -0.012113941,
        -0.062318724,
        0.030982977,
        0.03926206,
        -0.056871213,
        -0.13761872,
        -0.0014765395,
        -0.057043105,
        -0.13027647,
        -0.025552297,
        0.046032526,
        -0.05345821,
        0.036832996,
        -0.03738646,
        0.036666736,
        -0.11344009,
        0.44124857,
        -0.099996574,
        -0.05065545,
        0.010889863,
        -0.041492067,
        0.010356883,
        -0.09990315,
        0.021876248,
        0.035554506,
        0.012047599,
        0.032107882,
        0.0013247811,
        0.013760118,
        -0.043367345,
        0.035224173,
        -0.06647288,
        0.019304452,
        -0.075627916,
        0.03188215,
        0.04700672,
        -0.0035827374,
        -0.08004697,
        -0.037170198,
        0.06631083,
        0.0149377445,
        0.0057257386,
        0.06693812,
        0.037839465,
        -0.039044227,
        0.111429796,
        -0.013178063,
        -0.07920411,
        0.0083189625,
        0.018737413,
        0.029719241,
        -0.03135365,
        -0.040914457,
        0.06463453,
        -0.09628813,
        -0.019269977,
        -0.016202908,
        -0.0074832225,
        -0.027871331,
        -0.03084488,
        0.02567236,
        -0.029745758,
        -0.10102132,
        -0.04574,
        0.06521075,
        0.03343387,
        0.02120958,
        0.0037493987,
        0.07285586,
        -0.08761411,
        -0.055533823,
        0.05136137,
        0.019938383,
        -0.028352493,
        0.10602902,
        -0.009369838,
        0.10343905,
        -0.008510035,
        0.037497632,
        -0.06595699,
        -0.024218354,
        -0.0409845,
        0.067863524,
        -0.08990095,
        0.08478264,
        -0.06360712,
        -0.006948782,
        0.0500023,
        -0.01261503,
        0.024898859,
        -0.022857305,
        0.0034064366,
        -0.06228017,
        0.07527459,
        0.020483509,
        0.017716,
        0.021455886,
        -0.036116816,
        -0.6257064,
        0.0201812,
        -0.0325395,
        -0.008568142,
        -0.1607679,
        0.09092791,
        -0.019950947,
        -0.018901318,
        0.09484383,
        -0.016449891,
        0.048809897,
        -0.057875138,
        0.01620577,
        -0.0031058644,
        0.022759058,
        -0.006499859,
        0.033451132
      ]
    },
    {
      "id": "worldsfair-speaker-26",
      "conference": "worldsfair",
      "name": "Filip Makraduli",
      "role": "Machine Learning Engineer",
      "company": "Superlinked",
      "twitter": "https://x.com/f_makraduli",
      "talks": [
        {
          "title": "Turning My Obsidian Vault Into a Local AI Engineer",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        },
        {
          "title": "Weight Folding, CUDA Streams, and the Bug That Made My Model Speak Backwards",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0066849487,
        -0.009321641,
        0.052624997,
        -0.025786381,
        -0.024721164,
        -0.012195294,
        -0.041752383,
        -0.0046171267,
        0.06506175,
        -0.21984424,
        -0.07789203,
        -0.05590554,
        0.0097354725,
        -0.011312495,
        -0.033557955,
        -0.016035477,
        0.14251706,
        -0.11737196,
        0.0064466395,
        -0.13408053,
        -0.074042544,
        -0.08088654,
        -0.018112402,
        0.06539501,
        -0.030681327,
        0.080914594,
        -0.0046327366,
        0.030875113,
        -0.10932864,
        0.486331,
        -0.113570616,
        -0.025082402,
        0.029020693,
        -0.036116343,
        0.034848634,
        -0.060303077,
        -0.08380479,
        0.061906397,
        -0.07993383,
        0.08369443,
        0.026152715,
        0.01760525,
        -0.009162258,
        0.038121615,
        -0.017045733,
        -0.0038343102,
        -0.025757682,
        0.013338056,
        0.018571844,
        -0.020463526,
        -0.0612919,
        -0.015811952,
        0.12238374,
        0.004369702,
        -0.00877051,
        0.0141479345,
        0.07120975,
        -0.063895494,
        0.07875387,
        -0.06414317,
        -0.12185033,
        0.022746507,
        -0.008840358,
        0.09153206,
        -0.008672333,
        -0.053381756,
        0.05422981,
        -0.08965365,
        0.009111839,
        0.0009652453,
        0.07550658,
        -0.023376768,
        -0.044297356,
        -0.0057429704,
        -0.03063033,
        -0.032387745,
        -0.07130694,
        0.06558824,
        0.02360722,
        -0.006323568,
        0.0076983124,
        0.06309868,
        -0.029300043,
        -0.018273782,
        0.08715259,
        0.004774916,
        -0.042222936,
        0.043722827,
        -0.007787502,
        0.030128062,
        0.015344495,
        -0.0042950367,
        -0.094099596,
        -0.007978947,
        -0.03621179,
        0.06646394,
        -0.022932231,
        0.014441841,
        0.031158129,
        -0.008695441,
        0.015723849,
        -0.022295956,
        0.0027779585,
        -0.048838563,
        -0.016321713,
        -0.117866874,
        0.03844128,
        -0.051950406,
        0.0217422,
        -0.032353625,
        -0.074755095,
        -0.588721,
        0.011677685,
        -0.072008274,
        -0.026507832,
        -0.16728055,
        -0.010408599,
        0.012578406,
        0.02773383,
        0.0744613,
        0.033935256,
        0.0014892777,
        -0.03665517,
        0.031607006,
        0.06200668,
        -0.020966148,
        0.024207463,
        -0.023747852
      ]
    },
    {
      "id": "worldsfair-speaker-27",
      "conference": "worldsfair",
      "name": "Paul Bakaus",
      "role": "",
      "company": "Impeccable",
      "twitter": "",
      "talks": [
        {
          "title": "The Dark Arts of Skill Engineering",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "1:15pm-2:15pm",
          "type": "sponsor"
        },
        {
          "title": "Design at the Speed of Adjectives",
          "description": "Every design tool today operates at the wrong level of abstraction for AI-assisted engineering. Traditional tools give you padding sliders and color pickers, built for a world where designer and engineer are separate roles moving at separate speeds. Prompt-to-design tools one-shot a pretty landing page from a sentence, which is more dangerous because it looks like it's working. No serious design director hears a prompt and starts pushing pixels. The brief comes first. What's the emotional territory? What should this not feel like? Today's AI tools skip that discovery entirely. The result is output without intent. Technically competent, strategically empty. The right abstraction for a world where the designer is also the engineer lives between these extremes. Not pixels. Not prompts. Adjectives. \"Make it feel warmer.\" \"Strip it to its essence.\" \"Add tension.\" These are the controls a creative director actually thinks in. Drawing on lessons from building Impeccable, an open source design tool with 24 adjective-level commands like /bolder, /quieter, and /distill, I'll share what worked, what didn't, and how to apply this thinking to any AI interface where creative intent matters more than parameter control.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.024831474,
        0.0053003985,
        0.0305456,
        -0.014738784,
        0.0093006445,
        -0.056552786,
        -0.12128339,
        -0.083367296,
        0.10101129,
        -0.20043193,
        -0.040030602,
        -0.0014004103,
        0.024205074,
        -0.02035572,
        0.03609713,
        0.0103219235,
        0.053239588,
        -0.084361866,
        -0.0756553,
        -0.07630121,
        0.0165915,
        -0.08122442,
        -0.101423934,
        0.072919525,
        0.025935505,
        0.048076853,
        -0.026023977,
        0.025045367,
        -0.08394419,
        0.5027555,
        -0.054248754,
        0.07374487,
        0.0885703,
        -0.046086,
        0.027204517,
        -0.04899348,
        -0.060266625,
        0.026504595,
        -0.034142092,
        0.017602142,
        -0.10911629,
        0.031344853,
        -0.08509658,
        -0.000639487,
        -0.038502697,
        -0.06753596,
        -0.081464864,
        -0.031347685,
        0.011080841,
        0.0189023,
        0.022434656,
        0.04322101,
        -0.003742378,
        -0.053678002,
        0.05297169,
        0.01700601,
        0.014446631,
        -0.06265349,
        0.07464983,
        -0.10026431,
        -0.05160603,
        0.06709227,
        -0.0465924,
        0.04290903,
        0.045430887,
        -0.08457111,
        0.021873696,
        -0.09151527,
        0.038558435,
        -0.023776107,
        0.060302928,
        0.02791733,
        -0.060635705,
        -0.017775416,
        -0.071461305,
        -0.07238152,
        -0.05052614,
        0.038906455,
        -0.011686709,
        0.024702977,
        0.003088801,
        0.04201692,
        -0.024745846,
        -0.047908843,
        -0.013775696,
        0.021623997,
        -0.08654921,
        0.014351008,
        -0.025145743,
        0.03543476,
        0.0037074224,
        0.013311104,
        -0.058009915,
        0.019115778,
        -0.019548092,
        0.07389704,
        0.040507358,
        0.011761932,
        0.04833588,
        0.020892918,
        -0.058270905,
        -0.014497158,
        -0.019633695,
        -0.05167698,
        -0.038858227,
        -0.05511476,
        0.010902811,
        0.030992983,
        0.050381072,
        0.033450216,
        -0.04070554,
        -0.5978007,
        0.01235071,
        -0.04272725,
        -0.014417812,
        -0.15947321,
        0.05876727,
        -0.035936225,
        0.04867073,
        0.101291284,
        0.004035669,
        0.030530976,
        -0.05647725,
        -0.032449715,
        0.051738895,
        0.07184946,
        0.045199536,
        0.0442968
      ]
    },
    {
      "id": "worldsfair-speaker-28",
      "conference": "worldsfair",
      "name": "Daniel Han",
      "role": "",
      "company": "Unsloth",
      "twitter": "",
      "talks": [
        {
          "title": "Special topics in Kernels, RL, Reward Hacking in Agents",
          "description": "an advanced seminar (required prerequisites: Daniel's 2024 and 2025 hit AIE workshops)",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-5:30pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.034712516,
        0.028977206,
        0.042710934,
        0.036943205,
        0.0035580024,
        0.032495238,
        -0.0020588113,
        -0.028092105,
        0.025774622,
        -0.19872579,
        -0.037692454,
        -0.12893984,
        -0.08451009,
        0.034413498,
        -0.07146776,
        -0.034699213,
        0.093300015,
        -0.08688031,
        -0.075039975,
        -0.07845335,
        -0.003740943,
        -0.15696152,
        -0.022438865,
        0.09737043,
        -0.01087019,
        0.05203028,
        0.015639512,
        0.15023194,
        -0.059391014,
        0.43656322,
        -0.044885345,
        0.008450745,
        0.03939161,
        0.0059230584,
        0.073832035,
        -0.10345553,
        -0.0065173367,
        -0.024574706,
        0.020542605,
        0.03794173,
        -0.013375997,
        -0.032673303,
        0.034031555,
        0.02082258,
        -0.09624583,
        -0.040014416,
        -0.091420785,
        -0.04017842,
        -0.00018863923,
        -0.020537829,
        -0.035990514,
        0.018990763,
        0.07704385,
        -0.01630681,
        0.06339341,
        0.063131265,
        0.01078169,
        -0.05388278,
        0.107813396,
        0.056284487,
        -0.05318362,
        0.069388576,
        -0.024013648,
        0.009689204,
        -0.0009446291,
        -0.05141937,
        0.031453207,
        -0.049022987,
        0.040094335,
        0.0081749465,
        0.053819638,
        0.023094015,
        -0.0565642,
        0.01584258,
        -0.092731155,
        -0.040062837,
        -0.07054459,
        0.07684697,
        -0.057957415,
        0.04413186,
        -0.010538557,
        0.023326565,
        -0.08042376,
        0.0008277172,
        0.05081448,
        -0.033368822,
        -0.0070997076,
        -0.003738128,
        0.0022830123,
        0.012983029,
        0.08376401,
        0.050349593,
        -0.041693967,
        -0.0039483015,
        -0.040066477,
        0.0022275713,
        -0.034853514,
        0.08723113,
        -0.023484033,
        -0.040810432,
        0.017212832,
        0.014796011,
        0.08400931,
        -0.012924291,
        -0.05455164,
        -0.059329804,
        0.013895814,
        -0.029770767,
        0.031529315,
        -0.04229101,
        -0.0010907471,
        -0.64339226,
        -0.06544712,
        -0.040082656,
        -0.008503543,
        -0.111738965,
        0.042893786,
        -0.040260956,
        0.0778955,
        0.028797818,
        -0.020828424,
        0.0014367664,
        0.047581974,
        -0.0074117645,
        -0.09124215,
        0.006602664,
        0.034922298,
        0.0040141274
      ]
    },
    {
      "id": "worldsfair-speaker-29",
      "conference": "worldsfair",
      "name": "Justin Lebar",
      "role": "Software engineer",
      "company": "OpenXLA",
      "twitter": "https://x.com/lebar_dev",
      "talks": [
        {
          "title": "AI Infrastructure from the Ground Up",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.017052375,
        -0.009914227,
        -0.0003318347,
        0.051075745,
        0.045228608,
        0.009890959,
        -0.06725822,
        0.018670673,
        0.06349277,
        -0.21566148,
        -0.028890204,
        -0.07093293,
        0.018876005,
        -0.084870964,
        0.013479256,
        0.034857225,
        0.1711075,
        -0.056187168,
        -0.06765813,
        -0.09425345,
        0.03106943,
        -0.15020454,
        -0.027083948,
        0.07698303,
        -0.03377214,
        0.055386033,
        0.03411236,
        0.021002023,
        -0.0840811,
        0.4705092,
        -0.11780306,
        0.010380286,
        0.007888675,
        0.037224546,
        0.06819878,
        -0.08799601,
        -0.02310998,
        -0.011079953,
        0.011606149,
        0.013601448,
        0.01821603,
        -0.021951342,
        -0.039544974,
        0.013274187,
        -0.045685586,
        0.0073015383,
        -0.075860865,
        -0.057697926,
        -0.0039003522,
        0.0116640115,
        -0.033121113,
        0.014383215,
        0.028338013,
        -0.00023370708,
        0.008071894,
        0.008321106,
        -0.05543342,
        -0.06344771,
        0.09392259,
        0.030400896,
        -0.12408094,
        -0.033030808,
        -0.031280763,
        0.045486636,
        0.07060827,
        -0.057864625,
        0.0823165,
        -0.05026778,
        -0.054300502,
        0.0101606455,
        0.020138284,
        0.034670383,
        -0.03497869,
        0.013186728,
        -0.054070406,
        -0.016295265,
        -0.04761249,
        0.059682395,
        -0.0072367666,
        0.020544903,
        0.058384262,
        0.046407446,
        -0.10185802,
        0.051370278,
        0.07111983,
        -0.05022101,
        -0.022455445,
        0.07423099,
        -0.044490367,
        0.054432433,
        -0.003055109,
        -0.08197208,
        -0.009081997,
        -0.021720286,
        -0.029708276,
        0.100140005,
        -0.10075269,
        0.07233045,
        0.019582376,
        -0.022594135,
        -0.030012915,
        -0.09298027,
        0.06102185,
        0.031448364,
        -0.055997327,
        -0.08710748,
        0.016241724,
        -0.011719845,
        0.048283786,
        -0.07549563,
        -0.00071140705,
        -0.5787925,
        -0.013979369,
        -0.012443211,
        0.032370064,
        -0.1543331,
        0.037319236,
        -0.10331509,
        0.028499654,
        0.12087606,
        0.0024597305,
        -0.017157782,
        -0.021065095,
        0.05315529,
        0.034214053,
        0.0052371635,
        0.012823385,
        0.045838356
      ]
    },
    {
      "id": "worldsfair-speaker-30",
      "conference": "worldsfair",
      "name": "Michael Forrester",
      "role": "Principal Training Architect",
      "company": "Accenture",
      "twitter": "https://x.com/peopleforrester",
      "talks": [
        {
          "title": "Build a Platform, Unleash an Agent on it.... and Watch it Burn!",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.09718209,
        0.039085947,
        0.00550322,
        0.031399816,
        0.017859567,
        -0.00774888,
        -0.069198,
        0.023043394,
        0.020833721,
        -0.18795794,
        -0.07008114,
        -0.014571414,
        -0.01384347,
        -0.007243862,
        -0.0688167,
        -0.035179686,
        0.06664018,
        -0.0832623,
        -0.026395762,
        -0.062597506,
        -0.061314356,
        -0.15518951,
        -0.021341916,
        0.060634,
        -0.06357844,
        0.055406176,
        -0.0520015,
        0.037392583,
        -0.03194587,
        0.47735333,
        -0.07404801,
        0.028770452,
        0.11481287,
        0.0053179413,
        0.039841972,
        -0.037837766,
        -0.05206947,
        0.0117925955,
        -0.040328104,
        -0.014617634,
        -0.0740545,
        0.06748212,
        -0.007279836,
        -0.09175185,
        0.0000016969049,
        -0.017984204,
        -0.08246757,
        0.0023837576,
        0.035106625,
        -0.048013467,
        -0.08921527,
        0.0022300852,
        0.019737918,
        -0.08019608,
        0.06020158,
        0.016036872,
        0.030482313,
        -0.0009632959,
        0.12067846,
        -0.096895166,
        -0.03253732,
        -0.029346032,
        0.049592335,
        -0.019608602,
        -0.036234245,
        -0.023100536,
        0.092373885,
        0.027395073,
        0.02667508,
        -0.074198276,
        0.049930517,
        0.06276654,
        0.017158031,
        0.08138951,
        -0.017991984,
        -0.007741055,
        -0.09018005,
        0.07156263,
        -0.015092979,
        -0.035697784,
        0.016370188,
        0.061814006,
        -0.07771634,
        -0.023033304,
        -0.012534676,
        -0.09395404,
        -0.09416496,
        0.005322505,
        -0.015485961,
        0.05048156,
        0.01752751,
        0.03412496,
        -0.03485485,
        0.042024963,
        -0.035686597,
        0.01829489,
        -0.11110205,
        0.04338651,
        0.043574285,
        -0.03143768,
        -0.093229674,
        -0.062035095,
        0.032677926,
        -0.031007193,
        0.015127339,
        -0.08750356,
        -0.015770769,
        -0.01915548,
        0.08536084,
        0.008492446,
        -0.010508452,
        -0.57801354,
        -0.05740958,
        -0.054158337,
        0.05481744,
        -0.15088211,
        0.0919793,
        -0.015095659,
        0.009278724,
        0.11351166,
        0.07547969,
        0.048367932,
        -0.044061463,
        -0.0027504438,
        -0.028262494,
        -0.033941332,
        0.06619383,
        0.08642664
      ]
    },
    {
      "id": "worldsfair-speaker-31",
      "conference": "worldsfair",
      "name": "Louis-François Bouchard",
      "role": "Co-founder and CTO",
      "company": "Towards AI",
      "twitter": "https://x.com/Whats_AI",
      "talks": [
        {
          "title": "The AI Engineering Playbook: From Prototype to Production",
          "description": "",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.06156972,
        -0.005216698,
        0.034635425,
        0.02348175,
        -0.024574932,
        -0.022107331,
        -0.044191495,
        0.017206952,
        0.04329361,
        -0.19198859,
        -0.032221794,
        -0.074049264,
        -0.044306565,
        -0.007060431,
        -0.09793582,
        0.06062696,
        0.11595007,
        -0.0625735,
        0.0000068555096,
        -0.065344654,
        -0.103129946,
        -0.07031491,
        -0.010290384,
        0.047007952,
        -0.13815196,
        0.119769365,
        0.006576371,
        0.016462471,
        -0.08065266,
        0.4684454,
        -0.12239795,
        -0.041732613,
        0.048297387,
        -0.002964472,
        -0.0049719103,
        -0.03931822,
        -0.029949242,
        0.00016611462,
        0.024735194,
        0.022028415,
        -0.08449505,
        0.10271715,
        -0.008697079,
        0.014285423,
        -0.04593616,
        -0.029066691,
        -0.06928278,
        -0.016432824,
        -0.012555484,
        0.009361496,
        -0.066623494,
        0.04936596,
        -0.009678694,
        -0.075505935,
        0.063025825,
        0.002518574,
        0.06624741,
        -0.077475026,
        0.07621057,
        -0.069393136,
        -0.16862302,
        -0.12927689,
        0.004612883,
        0.06699768,
        -0.042212654,
        -0.0151095055,
        0.07755046,
        -0.059541658,
        -0.03832305,
        -0.0724607,
        0.04457419,
        -0.040707298,
        -0.054208558,
        -0.011283751,
        -0.0796648,
        -0.004208816,
        -0.012954088,
        0.0901795,
        0.06314812,
        -0.024354355,
        0.032137413,
        -0.000017051738,
        -0.044717237,
        -0.08117523,
        0.05859795,
        0.049139015,
        -0.024144122,
        0.06940573,
        0.01127076,
        0.07049109,
        0.044798966,
        -0.018523939,
        -0.06047979,
        0.027911834,
        -0.056921907,
        0.04058995,
        -0.011335173,
        0.042884782,
        0.047095627,
        -0.0660949,
        -0.015696125,
        -0.020370957,
        0.0965641,
        -0.050414596,
        -0.014390191,
        -0.069622576,
        0.008544146,
        -0.048856582,
        0.03556791,
        -0.0075186873,
        -0.054327462,
        -0.5796481,
        -0.035751715,
        -0.020028641,
        0.008261664,
        -0.13971353,
        0.04636904,
        -0.003524636,
        -0.041028485,
        0.041056793,
        -0.04167439,
        -0.017974146,
        0.0027704916,
        0.022013808,
        0.07441886,
        0.077782474,
        0.055520594,
        0.019510776
      ]
    },
    {
      "id": "worldsfair-speaker-32",
      "conference": "worldsfair",
      "name": "Emile Baizel",
      "role": "",
      "company": "Amazon AGI Lab",
      "twitter": "",
      "talks": [
        {
          "title": "Build with Perception Agents",
          "description": "Human-agent collaboration is changing, becoming more visual. Models can perceive, point, and verify, but most agents still rely on us typing a paragraph to explain what we're looking at. Meet perception agents: agents that see what you see, verify their own work, and let you point, draw, and describe instead of just typing.",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06427282,
        0.040690105,
        0.033733286,
        0.04070905,
        -0.04296244,
        -0.01339607,
        -0.052261565,
        -0.030770874,
        -0.03410537,
        -0.17568398,
        0.025305647,
        -0.07467694,
        0.026015373,
        -0.0055575236,
        -0.06270547,
        -0.065727435,
        0.13411306,
        -0.12831172,
        -0.062096458,
        -0.03110966,
        -0.03234309,
        -0.12373681,
        -0.04203348,
        0.007072022,
        -0.010756512,
        0.10100068,
        -0.017516661,
        0.056473393,
        -0.0823513,
        0.5030083,
        -0.08250352,
        -0.044068135,
        0.005390535,
        -0.030357543,
        0.0384053,
        -0.06689935,
        0.09115268,
        0.011388371,
        0.059757046,
        -0.019184561,
        -0.008549035,
        0.01860932,
        -0.025911301,
        -0.006275854,
        -0.010849071,
        0.029645054,
        -0.011604298,
        -0.0310191,
        -0.0071941153,
        0.005551907,
        -0.015845388,
        0.0059715095,
        0.0500807,
        -0.044616047,
        0.0689904,
        0.030296484,
        0.0383248,
        -0.04701175,
        0.05549656,
        0.034752894,
        0.012157006,
        -0.09860721,
        -0.035636965,
        0.052860223,
        0.034255225,
        -0.018679949,
        0.022451172,
        -0.07744818,
        0.0380646,
        -0.081253886,
        0.030229209,
        -0.011422756,
        -0.05720387,
        0.077335544,
        -0.05752694,
        -0.08827724,
        -0.06552234,
        0.012418235,
        -0.03687368,
        0.027433416,
        0.015607909,
        0.04095191,
        -0.09044197,
        -0.043649435,
        0.0050726137,
        -0.026355257,
        -0.06862259,
        0.03592953,
        -0.058477063,
        -0.030511847,
        0.037747625,
        0.00456046,
        -0.015527767,
        0.0028247866,
        -0.051933482,
        0.002915993,
        -0.05731866,
        0.017282479,
        0.0605529,
        0.027999269,
        -0.023340466,
        -0.043937705,
        0.030594677,
        0.010072999,
        0.009120444,
        -0.023608845,
        0.048939656,
        0.051518653,
        0.064071916,
        -0.084449746,
        0.03336944,
        -0.636589,
        -0.044633005,
        0.0031076858,
        0.004273906,
        -0.16206639,
        0.025667919,
        -0.05438616,
        0.026968969,
        0.032044984,
        0.00085778505,
        0.022065992,
        -0.03813697,
        -0.0128114885,
        -0.0059192535,
        0.016590057,
        0.032361936,
        -0.016928434
      ]
    },
    {
      "id": "worldsfair-speaker-33",
      "conference": "worldsfair",
      "name": "Shruti Arora",
      "role": "",
      "company": "Amazon AGI Lab",
      "twitter": "",
      "talks": [
        {
          "title": "Build with Perception Agents",
          "description": "Human-agent collaboration is changing, becoming more visual. Models can perceive, point, and verify, but most agents still rely on us typing a paragraph to explain what we're looking at. Meet perception agents: agents that see what you see, verify their own work, and let you point, draw, and describe instead of just typing.",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.027449757,
        0.10623232,
        0.026813133,
        0.04038209,
        0.004692984,
        -0.061489575,
        -0.06351052,
        -0.009749098,
        0.007195482,
        -0.18163706,
        -0.0067826803,
        -0.04498287,
        0.054804824,
        -0.03716905,
        -0.040848598,
        -0.054729335,
        0.09737328,
        -0.06305482,
        -0.089228034,
        -0.06797352,
        0.00040059767,
        -0.10922338,
        -0.04976526,
        0.06736029,
        -0.032907777,
        0.12627636,
        0.034730315,
        0.017201468,
        -0.11101832,
        0.47954428,
        -0.08115466,
        -0.04201924,
        0.02102268,
        -0.060216483,
        0.06565505,
        -0.066017,
        0.033750545,
        -0.030584056,
        0.073747195,
        0.008146619,
        -0.0139317205,
        0.015703117,
        -0.036669996,
        0.025435947,
        -0.04779376,
        -0.03121918,
        -0.04796893,
        0.0013652038,
        -0.0022391712,
        0.02008868,
        -0.0295838,
        -0.012145464,
        0.030802958,
        -0.045794576,
        0.04555149,
        0.08734942,
        0.012628255,
        -0.024155524,
        0.08167321,
        0.029071072,
        -0.04407294,
        -0.05021141,
        -0.019257603,
        0.076876946,
        -0.008982005,
        -0.06609681,
        0.06530475,
        -0.072492026,
        0.019307977,
        -0.06690057,
        -0.006985723,
        0.01265283,
        -0.003206136,
        0.06935846,
        0.008192981,
        -0.12163024,
        -0.05279331,
        0.04471743,
        0.0057234312,
        0.034511976,
        -0.027851319,
        -0.0047986424,
        -0.107174754,
        -0.04441501,
        0.012997933,
        0.0002531857,
        -0.059426613,
        0.035592426,
        -0.06158735,
        0.006124745,
        0.010025779,
        0.030607974,
        -0.032941293,
        0.012873176,
        -0.06487762,
        -0.0046996675,
        -0.054447506,
        0.04280939,
        0.03756216,
        0.017738033,
        0.021445772,
        -0.018491605,
        -0.0057836776,
        0.02978293,
        -0.035312798,
        -0.018445138,
        0.060295258,
        0.022239495,
        0.09607647,
        -0.0772253,
        -0.010125797,
        -0.63795245,
        0.021843959,
        -0.024430897,
        -0.005377017,
        -0.16965635,
        0.018583117,
        -0.06494036,
        0.04742073,
        0.028736467,
        -0.009091743,
        0.07150962,
        -0.033303566,
        -0.030039478,
        -0.026762836,
        0.0018616746,
        0.042508446,
        0.008571668
      ]
    },
    {
      "id": "worldsfair-speaker-34",
      "conference": "worldsfair",
      "name": "Zhengyao Jiang",
      "role": "Co-founder and CEO",
      "company": "Weco AI",
      "twitter": "https://x.com/zhengyaojiang",
      "talks": [
        {
          "title": "Hands-on AutoResearch: Cracking OpenAI's Parameter Golf",
          "description": "Heard about autoresearch, or tried it a few times in playground settings? This hands-on tutorial teaches you how to use autoresearch on one of the most serious challenges in ML this year: OpenAI's Parameter Golf.\n\nThe challenge: train the best language model that fits in just 16MB. We entered our autoresearch agent this past spring, and it outperformed the field of over 1,000 participants. You'll learn how we approached it, then get to do it yourself: kick off an autoresearch agent, watch it improve a tiny language model's training script, steer it when progress stalls, and visualize your results. You'll leave with a working autoresearch setup you can point at your own code.",
          "day": "Day 1 — Workshop Day",
          "time": "2:20pm-4:20pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.037473843,
        0.11157338,
        0.014731531,
        0.029645877,
        -0.030112665,
        -0.0343172,
        -0.048556466,
        0.014811264,
        0.016043846,
        -0.25930908,
        -0.03974314,
        0.014062076,
        0.0051100384,
        0.029979857,
        -0.049025737,
        -0.026482385,
        0.03217424,
        -0.09574593,
        -0.05765259,
        -0.027756615,
        -0.09028842,
        -0.11382969,
        -0.044693936,
        0.089118615,
        -0.050824,
        0.07402358,
        0.039019022,
        -0.014428235,
        -0.09473624,
        0.4067415,
        -0.10139728,
        -0.0006962688,
        0.06738289,
        -0.010727767,
        0.0466687,
        -0.0653566,
        0.010355547,
        -0.031569652,
        -0.023156743,
        0.020161152,
        -0.04135957,
        0.044275902,
        -0.019265238,
        0.0746861,
        -0.021332467,
        -0.0035852592,
        -0.097259715,
        0.02246192,
        0.044457957,
        -0.05270475,
        -0.04110793,
        -0.00800902,
        0.029673088,
        0.02701437,
        0.034634374,
        0.004919841,
        -0.029895721,
        -0.064446695,
        0.1262499,
        -0.009435054,
        -0.09818102,
        0.001269912,
        0.0021875254,
        0.03818323,
        -0.05506905,
        -0.008052946,
        -0.053131003,
        -0.035554282,
        0.045982517,
        -0.051251,
        0.044426337,
        0.0062179198,
        -0.028426163,
        0.025622873,
        -0.1402821,
        -0.09686608,
        -0.056640264,
        0.07291904,
        0.00028790478,
        0.025370806,
        -0.01191404,
        0.040381067,
        -0.061848152,
        -0.06471417,
        0.052319698,
        0.032727737,
        -0.08268164,
        0.05341248,
        -0.07611317,
        0.05342378,
        0.058303773,
        -0.020208467,
        -0.06952237,
        0.036158565,
        -0.11554222,
        0.059588864,
        -0.0068563437,
        0.0039269547,
        0.032687172,
        -0.0097803725,
        0.0017138568,
        0.006088458,
        0.09924088,
        -0.023157086,
        -0.07222318,
        -0.07211418,
        0.007941368,
        -0.03869727,
        -0.011515679,
        -0.01668379,
        -0.06862756,
        -0.6391743,
        0.011336434,
        -0.03167611,
        -0.006003504,
        -0.13013026,
        0.083992586,
        -0.02053777,
        -0.0055397335,
        0.049929045,
        -0.06444412,
        0.042830408,
        0.011632589,
        0.033896547,
        -0.057464924,
        0.0572025,
        0.008150277,
        0.02834463
      ]
    },
    {
      "id": "worldsfair-speaker-35",
      "conference": "worldsfair",
      "name": "Ang Li",
      "role": "CEO and Co-Founder",
      "company": "Simular",
      "twitter": "https://x.com/angli_ai",
      "talks": [
        {
          "title": "Reliable Computer Use Agents require coding",
          "description": "Even the world's best computer-use agents cannot repeat their successes at the moment. Agents that write code — emitting structured selector-based actions instead of clicking pixels — break through that ceiling. We'll share two years of experience from Simular's production agent platform, the architectural decisions that mattered (refs over pixels, code as substrate, Simulang DSL), and a live demo: a 30-step unattended Windows workflow, side-by-side with a vision-only baseline. If you're shipping agents to real users, this is the playbook.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.039218053,
        0.02474505,
        -0.012292754,
        -0.0039947713,
        -0.036553282,
        0.012624454,
        0.013584545,
        -0.034316514,
        -0.000042344887,
        -0.24182557,
        -0.010256232,
        0.0013862213,
        0.0052086813,
        0.032794144,
        -0.042275175,
        0.016587924,
        0.086446166,
        -0.12760249,
        -0.033714324,
        -0.09144477,
        -0.06361648,
        -0.018377995,
        0.019861108,
        0.032873787,
        0.0021174904,
        0.066978656,
        -0.0032154205,
        0.051320717,
        -0.09545671,
        0.47363162,
        -0.07015815,
        0.019223444,
        0.031632386,
        -0.090330206,
        0.03301074,
        -0.057832483,
        -0.058796026,
        -0.0071973624,
        0.03460441,
        0.03912893,
        -0.02567982,
        0.021052167,
        -0.02003432,
        0.05481139,
        -0.10726217,
        0.011959786,
        -0.007338424,
        0.011086363,
        -0.01095581,
        -0.011063124,
        -0.024259422,
        -0.03401595,
        0.008221579,
        -0.0479051,
        0.014491069,
        0.009260199,
        0.009182182,
        -0.0792171,
        0.09467791,
        0.028821442,
        -0.10442071,
        -0.007188835,
        0.048245266,
        0.059221286,
        0.044994794,
        -0.091572635,
        0.031728413,
        -0.06578747,
        0.06442743,
        -0.07045496,
        0.03492393,
        0.051259596,
        -0.039333705,
        0.035836354,
        -0.10514086,
        -0.10404983,
        -0.052869897,
        0.07982775,
        -0.04397171,
        -0.01900434,
        -0.027563738,
        0.08193596,
        -0.08535864,
        -0.04725019,
        0.018055411,
        0.012757171,
        -0.035640325,
        0.0202063,
        -0.04916689,
        0.06596341,
        0.007557374,
        0.022202836,
        -0.052969635,
        -0.0017625892,
        -0.10066645,
        0.031429607,
        -0.07395366,
        0.02589055,
        -0.036554553,
        0.0176022,
        0.010533973,
        0.013991993,
        0.08602192,
        0.0002348457,
        -0.04514239,
        -0.081449494,
        -0.029350027,
        -0.020932173,
        0.047663353,
        -0.026804367,
        -0.028877907,
        -0.6225158,
        0.023145193,
        -0.012359244,
        -0.06898585,
        -0.15001774,
        0.07255269,
        -0.033856917,
        0.04496339,
        0.040459223,
        -0.046036102,
        0.06661764,
        -0.025893826,
        0.01558545,
        -0.043034453,
        -0.013020627,
        0.05219184,
        0.044685222
      ]
    },
    {
      "id": "worldsfair-speaker-36",
      "conference": "worldsfair",
      "name": "Shubhankar Srivastava",
      "role": "AI Engineer",
      "company": "Browserbase",
      "twitter": "https://x.com/_shubhankar",
      "talks": [
        {
          "title": "Hill-climbing Skills: How to Improve Agents Without Touching the Model",
          "description": "Agent Capability is now highly dependent on the markdown files read at runtime -- skills.This workshop treats skills as a first-class optimization surface. We borrow the concept of autoresearch (from Karpathy) and apply it to the skills your agents already read. You'll see how we at Browserbase did the same for browser agents, enabling our customers to scale the coverage of their browser agents while improving performance(2x faster runs) and optimizing for token spend(upto 10x cheaper).You'll leave with a working http://SKILL.md you generated through an auto-research loop, and a mental model for when skill optimization beats fine-tuning or prompt engineering.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.062733024,
        0.07193454,
        0.05628085,
        0.009404838,
        -0.015587365,
        -0.0053753555,
        -0.12857738,
        -0.02101301,
        0.011870316,
        -0.2376248,
        -0.050417166,
        0.017681617,
        0.014541185,
        -0.023063598,
        -0.031699248,
        -0.036067627,
        0.098807946,
        -0.04975193,
        -0.045908555,
        -0.07356356,
        -0.06353669,
        -0.061855875,
        -0.0013001673,
        0.067718126,
        -0.030325571,
        0.12641796,
        0.04969087,
        0.03926453,
        -0.06952283,
        0.4755671,
        -0.10644367,
        0.011103109,
        0.06740237,
        0.013687407,
        0.018227875,
        -0.07077942,
        -0.050029885,
        -0.008203048,
        0.0058987914,
        -0.000006514745,
        -0.04757233,
        -0.019829521,
        -0.02429971,
        0.0108134085,
        -0.026154438,
        0.020926982,
        -0.029981386,
        0.026065446,
        -0.02348449,
        -0.023392754,
        -0.056733858,
        -0.0044178795,
        0.09408052,
        -0.018899798,
        0.04298202,
        0.043444257,
        0.049854636,
        -0.04098856,
        0.13585995,
        0.01759672,
        -0.06615302,
        0.026000664,
        0.027388781,
        -0.0057010725,
        0.018069642,
        -0.06985363,
        0.081070036,
        -0.06533218,
        0.034547877,
        -0.0022684284,
        0.050284058,
        0.0034088518,
        -0.022201506,
        0.06513252,
        -0.08446854,
        -0.1039716,
        -0.06772492,
        0.023576422,
        0.035711043,
        0.025965938,
        -0.050181545,
        0.014428016,
        -0.09428906,
        -0.037873153,
        0.03122261,
        -0.026705608,
        -0.026730726,
        0.0025630028,
        0.002313194,
        0.07880546,
        0.016389664,
        0.015097341,
        -0.09602815,
        0.0072478284,
        -0.051926237,
        -0.007984649,
        -0.031390667,
        0.07205258,
        0.040506296,
        0.009700787,
        0.0349904,
        -0.0107466,
        0.057629563,
        -0.043034427,
        0.029809019,
        -0.018694527,
        0.026208673,
        -0.029650265,
        0.0062369355,
        -0.008397018,
        -0.034914345,
        -0.6389165,
        0.01722104,
        0.007376242,
        -0.043507453,
        -0.12671809,
        0.02762902,
        0.0033152048,
        0.015031057,
        0.056173723,
        -0.018794663,
        0.06826133,
        -0.038681522,
        -0.027162317,
        -0.039659567,
        0.025759628,
        0.0358296,
        0.06130016
      ]
    },
    {
      "id": "worldsfair-speaker-37",
      "conference": "worldsfair",
      "name": "Bereket Engida",
      "role": "Founder & CEO",
      "company": "Better Auth",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Auth",
          "description": "Better Auth has grown to 27k GitHub stars and over 1.5M weekly downloads, becoming a popular choice for developers who want to own their authentication stack. We recently introduced Agent Auth, a protocol designed to support autonomous and delegated agents operating services for an organization or a user. It allows agents to dynamically negotiate capabilities, manage access boundaries, and maintain secure authorization flows. This session will break down the protocol design and demonstrate it live, showing how agents can securely authenticate and operate with dynamic permissions.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.08499882,
        0.014568161,
        0.055756688,
        -0.058920946,
        -0.026806362,
        -0.015693748,
        -0.112989135,
        -0.00513931,
        0.04067231,
        -0.21626939,
        -0.053211123,
        -0.064464316,
        -0.019872904,
        -0.04124776,
        -0.032789018,
        -0.0399029,
        0.041573796,
        -0.092649005,
        -0.026683217,
        -0.057228673,
        -0.035658136,
        -0.071395166,
        0.032513388,
        0.047838546,
        -0.048809864,
        0.09102238,
        -0.06002589,
        0.070787236,
        -0.05789878,
        0.46197805,
        -0.03751459,
        -0.026835887,
        -0.013477564,
        -0.033141144,
        0.025763998,
        -0.057284165,
        -0.0019791035,
        0.005473296,
        0.04431979,
        0.051934768,
        -0.05563854,
        0.05327212,
        -0.06353697,
        0.04020269,
        0.021337697,
        0.012705951,
        -0.06820746,
        -0.0022674005,
        -0.0062975953,
        0.013800667,
        -0.026302377,
        0.007823438,
        0.14115721,
        -0.020877348,
        0.048367843,
        0.022194324,
        0.045437425,
        -0.051662236,
        0.10993336,
        0.0075623565,
        -0.06315373,
        -0.0039189914,
        -0.009901122,
        0.061374456,
        0.005152925,
        -0.09225686,
        0.02787244,
        -0.016291736,
        -0.016498484,
        -0.052004326,
        0.017923681,
        0.1068972,
        -0.04235483,
        0.018675668,
        -0.067001775,
        -0.07951958,
        -0.01264398,
        0.041375037,
        -0.017915243,
        0.02213017,
        -0.0073229354,
        0.026614152,
        -0.10178018,
        -0.02423111,
        0.04946279,
        -0.026964765,
        -0.036644865,
        0.05565378,
        0.039065797,
        0.04171508,
        0.006583358,
        0.044851217,
        -0.07568383,
        0.04261538,
        -0.06083938,
        0.05606386,
        -0.08061676,
        -0.017100817,
        0.026094934,
        -0.00022892175,
        -0.0051489053,
        -0.037657868,
        0.10327515,
        -0.041447587,
        0.02794296,
        0.015044419,
        -0.008667184,
        0.012353101,
        0.10559351,
        0.014006038,
        -0.030739188,
        -0.6305747,
        -0.039140847,
        0.033041053,
        -0.03580011,
        -0.16337794,
        0.045009516,
        0.013037883,
        0.071872376,
        0.0833251,
        -0.076489426,
        0.049214657,
        -0.009881379,
        0.0579596,
        -0.032800663,
        0.021594914,
        -0.06619853,
        -0.012771142
      ]
    },
    {
      "id": "worldsfair-speaker-38",
      "conference": "worldsfair",
      "name": "Paola Estefania",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Auth",
          "description": "Better Auth has grown to 27k GitHub stars and over 1.5M weekly downloads, becoming a popular choice for developers who want to own their authentication stack. We recently introduced Agent Auth, a protocol designed to support autonomous and delegated agents operating services for an organization or a user. It allows agents to dynamically negotiate capabilities, manage access boundaries, and maintain secure authorization flows. This session will break down the protocol design and demonstrate it live, showing how agents can securely authenticate and operate with dynamic permissions.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.09408742,
        0.05604738,
        0.019609088,
        -0.052067358,
        0.04075393,
        -0.01624767,
        -0.07565853,
        -0.057587173,
        0.03659406,
        -0.18609639,
        -0.014028864,
        -0.050350226,
        -0.048318937,
        -0.06315268,
        -0.037207752,
        0.0032963208,
        0.03649812,
        -0.11704789,
        -0.06481468,
        -0.03275499,
        -0.073230214,
        -0.12900786,
        -0.017123094,
        0.07840506,
        -0.03019814,
        0.0678553,
        -0.045456212,
        0.0504486,
        -0.07920956,
        0.47063774,
        -0.02273325,
        -0.008592631,
        0.020571453,
        -0.034707572,
        0.024934331,
        -0.093713544,
        -0.014632704,
        -0.005652842,
        0.0710551,
        -0.0061721075,
        -0.065341584,
        0.055203084,
        -0.04824133,
        0.029143,
        -0.007479684,
        -0.004211816,
        -0.07670299,
        -0.001314176,
        -0.0035632832,
        0.009958011,
        -0.023997176,
        0.030833106,
        0.07772228,
        -0.00013464491,
        0.045232642,
        -0.006091472,
        -0.01001751,
        -0.022130119,
        0.09545722,
        -0.03933386,
        -0.021897472,
        -0.021604087,
        -0.034355775,
        0.04737893,
        0.047221888,
        -0.0715417,
        0.08109065,
        -0.01578064,
        0.054388784,
        -0.11607648,
        0.065997764,
        0.03225892,
        0.005074074,
        0.018335909,
        -0.053073626,
        -0.14005564,
        -0.05728303,
        0.024446808,
        0.03196179,
        0.070763834,
        -0.057525296,
        -0.02639207,
        -0.1298901,
        -0.05085085,
        0.05463554,
        0.030421913,
        -0.05146058,
        0.05807526,
        -0.033621132,
        0.033953283,
        0.042776044,
        0.08408167,
        -0.06089403,
        0.029944621,
        -0.09047355,
        0.06477815,
        -0.027456796,
        0.030920941,
        0.008267771,
        0.004572897,
        0.011057275,
        -0.0947637,
        0.07440084,
        -0.036116306,
        0.045022875,
        -0.015542861,
        -0.049683943,
        0.002411287,
        0.095215835,
        0.03840136,
        0.038418304,
        -0.6012209,
        -0.038257644,
        0.01801473,
        -0.03081191,
        -0.16336982,
        0.058243252,
        -0.030299485,
        0.06828626,
        0.085263886,
        -0.042541746,
        0.06093849,
        -0.053591166,
        -0.011432862,
        -0.032291114,
        0.011244065,
        -0.020337239,
        -0.0080958735
      ]
    },
    {
      "id": "worldsfair-speaker-39",
      "conference": "worldsfair",
      "name": "Mark Lummus",
      "role": "",
      "company": "PayPal",
      "twitter": "",
      "talks": [
        {
          "title": "Burn your flags: How PayPal designs interactive CLI tools for agents",
          "description": "The common guidance for designing complex CLI tooling that agents can use is to add a \"non-interactive\" mode, where a normally interactive and flow-based command can be executed in a single pass by feeding it a bunch of flags. This is necessary for deterministic automation, but agents aren't scripts; they aren't really constrained in the same way, and they benefit greatly from the same step-by-step contextual workflows that humans do. The problem is that most agents (Codex excepted) cannot interactively drive a prompt-based CLI out of the box. In this workshop, PayPal goes deep on some techniques we've used in our upcoming `paypal` CLI that you can steal to make your complex CLI workflow tool agent-usable — without giving up the guardrails and guidance that interactive CLI tools provide. At the end of the workshop, you'll understand the benefits, patterns, and pitfalls of exposing a truly interactive mode for your CLI tool to agents.\n\nSummary: Build agent-friendly CLI tools with interactive workflows, guardrails, and practical design patterns for real-world automation.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        -0.015724596,
        0.047827404,
        0.025103979,
        -0.01404746,
        0.02322647,
        -0.062703766,
        -0.00058036443,
        -0.048824407,
        0.04125533,
        -0.23203714,
        -0.016916428,
        -0.09492731,
        -0.02088863,
        -0.03436912,
        -0.015917478,
        -0.0044017,
        0.120688334,
        -0.09657324,
        -0.11496716,
        -0.04686041,
        -0.04647611,
        -0.08733055,
        -0.037019823,
        0.051504124,
        0.026551928,
        0.08354812,
        0.016542105,
        0.010556321,
        -0.045468632,
        0.4620419,
        -0.09051684,
        0.08308342,
        0.045668315,
        0.0029430566,
        0.020865073,
        -0.020608759,
        -0.017152127,
        0.0056216847,
        -0.006317091,
        -0.021623002,
        0.022368813,
        0.020746497,
        -0.01687061,
        -0.025605965,
        -0.05520653,
        0.0034993985,
        -0.046016067,
        -0.05827893,
        0.0042067776,
        -0.040603332,
        -0.018295638,
        -0.013550826,
        0.04458358,
        0.036449995,
        0.086871564,
        -0.024880784,
        0.024913214,
        -0.029573169,
        0.0999439,
        0.03334113,
        -0.074742064,
        0.033493605,
        -0.06056301,
        0.026696298,
        -0.02983848,
        -0.023650492,
        0.0646549,
        -0.03670719,
        0.06743787,
        -0.008405766,
        0.042711683,
        0.052538164,
        -0.101370975,
        0.006175877,
        -0.038036365,
        -0.09645919,
        -0.054033432,
        0.05192359,
        -0.017302614,
        0.02555521,
        0.03368898,
        0.038207237,
        -0.053388007,
        -0.07733056,
        0.012579172,
        0.019359142,
        0.018940276,
        0.06271816,
        -0.02229751,
        -0.052296877,
        -0.00090072915,
        0.0007795082,
        0.0029558192,
        -0.025312092,
        -0.099229574,
        0.032758217,
        -0.095782556,
        0.00046578152,
        -0.039197978,
        0.04172124,
        -0.10399407,
        -0.034335583,
        0.03593464,
        -0.053310364,
        0.013402062,
        -0.07601748,
        -0.000038562383,
        -0.014519096,
        0.008649208,
        -0.060044218,
        -0.039206367,
        -0.64209867,
        -0.062352907,
        -0.027445251,
        0.054560933,
        -0.13927138,
        0.10157025,
        -0.00261235,
        0.008387426,
        0.063593544,
        -0.030690994,
        0.00990867,
        -0.044703912,
        -0.015655667,
        -0.022731613,
        -0.004527503,
        0.07034354,
        -0.04787519
      ]
    },
    {
      "id": "worldsfair-speaker-40",
      "conference": "worldsfair",
      "name": "Navinkumar Patil",
      "role": "Staff Software Engineer",
      "company": "PayPal",
      "twitter": "",
      "talks": [
        {
          "title": "Burn your flags: How PayPal designs interactive CLI tools for agents",
          "description": "The common guidance for designing complex CLI tooling that agents can use is to add a \"non-interactive\" mode, where a normally interactive and flow-based command can be executed in a single pass by feeding it a bunch of flags. This is necessary for deterministic automation, but agents aren't scripts; they aren't really constrained in the same way, and they benefit greatly from the same step-by-step contextual workflows that humans do. The problem is that most agents (Codex excepted) cannot interactively drive a prompt-based CLI out of the box. In this workshop, PayPal goes deep on some techniques we've used in our upcoming `paypal` CLI that you can steal to make your complex CLI workflow tool agent-usable — without giving up the guardrails and guidance that interactive CLI tools provide. At the end of the workshop, you'll understand the benefits, patterns, and pitfalls of exposing a truly interactive mode for your CLI tool to agents.\n\nSummary: Build agent-friendly CLI tools with interactive workflows, guardrails, and practical design patterns for real-world automation.",
          "day": "Day 1 — Workshop Day",
          "time": "4:30pm-5:30pm",
          "type": "workshop"
        }
      ],
      "embedding": [
        0.004182023,
        0.053304866,
        0.048612647,
        0.015501476,
        0.0361143,
        -0.024269698,
        -0.017743744,
        -0.009364662,
        -0.0045700762,
        -0.21113293,
        0.0013152019,
        -0.062411815,
        -0.025708126,
        -0.034301784,
        -0.019094443,
        -0.077330925,
        0.11044086,
        -0.112709165,
        -0.08811224,
        -0.05045537,
        -0.093213566,
        -0.10763662,
        -0.05158604,
        0.0529195,
        0.02466734,
        0.04301208,
        -0.012910165,
        0.027953641,
        -0.050596558,
        0.48742592,
        -0.096110255,
        0.04287657,
        0.10335215,
        -0.02356026,
        0.015976977,
        -0.033993892,
        0.045608036,
        -0.0010000862,
        -0.011483017,
        -0.05969932,
        0.039837457,
        0.008891618,
        -0.045431335,
        -0.0015626486,
        -0.07609677,
        -0.001085626,
        -0.09374709,
        -0.032555122,
        0.0038516645,
        -0.044957142,
        0.0031690456,
        -0.02459195,
        0.034741204,
        0.0038288762,
        0.03088592,
        -0.0034140267,
        -0.0018385292,
        -0.09827134,
        0.08197057,
        -0.0057872664,
        -0.062956005,
        0.058200363,
        -0.01726781,
        0.03524615,
        0.00026804185,
        -0.04136075,
        0.08063868,
        -0.053615566,
        0.028427636,
        0.028073344,
        0.08582718,
        -0.00057457713,
        -0.058496155,
        0.050405245,
        -0.051912647,
        -0.096302725,
        -0.05148393,
        0.05833881,
        0.020376597,
        0.011942862,
        0.029384574,
        0.048170656,
        -0.09149008,
        -0.05969812,
        0.02778839,
        -0.00882655,
        0.0013974025,
        0.027610997,
        -0.027733596,
        -0.0089308955,
        -0.0032592602,
        -0.03625732,
        -0.0027640373,
        -0.039905302,
        -0.14544658,
        0.04619224,
        -0.06231373,
        0.0056336247,
        -0.02193196,
        0.076374024,
        -0.03408503,
        -0.04201881,
        0.033633504,
        -0.062834516,
        -0.037620977,
        -0.07365147,
        0.000084686,
        -0.031615485,
        0.022359885,
        -0.03280053,
        -0.05568764,
        -0.6105612,
        -0.0031390407,
        -0.039750855,
        0.0054170773,
        -0.13647066,
        0.13990448,
        -0.007622485,
        0.05238409,
        0.08679581,
        -0.047993533,
        0.051830478,
        -0.038428187,
        -0.010492174,
        -0.0320662,
        -0.045220964,
        0.016483305,
        -0.0072156494
      ]
    },
    {
      "id": "worldsfair-speaker-41",
      "conference": "worldsfair",
      "name": "Shawn Wang",
      "role": "Founder & Editor",
      "company": "Latent Space",
      "twitter": "https://x.com/swyx",
      "talks": [
        {
          "title": "swyx keynote and snyk track intro",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "9:00am-9:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.013206693,
        0.025543641,
        -0.0008473797,
        -0.00554576,
        0.002409723,
        -0.03400307,
        -0.050705012,
        0.014743135,
        0.08775267,
        -0.21840978,
        -0.031473797,
        -0.06828454,
        -0.0024616192,
        -0.025413401,
        -0.10274374,
        0.020288061,
        0.07813876,
        -0.11759472,
        -0.06662469,
        -0.03351968,
        -0.007445452,
        -0.012878716,
        0.0060689733,
        0.06973735,
        -0.04619184,
        0.05903148,
        0.00704849,
        0.066987135,
        -0.12991573,
        0.4886766,
        0.00110716,
        -0.0063193967,
        0.06194432,
        0.005066051,
        0.0346313,
        -0.077386774,
        -0.020876005,
        -0.01947518,
        -0.033807714,
        0.048577175,
        0.040825866,
        -0.003852515,
        0.012087962,
        0.021546243,
        -0.0756486,
        0.02788852,
        -0.09386838,
        -0.022860913,
        -0.00994024,
        0.04143602,
        -0.089959554,
        -0.0010465309,
        0.057382,
        -0.042197857,
        -0.014099835,
        0.05209109,
        -0.010977695,
        -0.06148568,
        -0.013215011,
        -0.023468647,
        -0.1452578,
        -0.016830634,
        0.045595396,
        0.06677879,
        -0.04753406,
        -0.094620004,
        0.02344179,
        -0.043026827,
        -0.040388096,
        -0.01857211,
        0.029620137,
        -0.020057615,
        -0.053317342,
        0.079872735,
        -0.116354145,
        -0.055956617,
        0.02471288,
        0.05782436,
        -0.018263081,
        -0.018342232,
        0.023986971,
        0.002129177,
        -0.08028387,
        -0.09542429,
        0.0011708104,
        -0.0497877,
        -0.032634225,
        0.041224018,
        -0.019557975,
        0.0324048,
        0.0497199,
        -0.0053749103,
        -0.051053364,
        0.010387693,
        -0.06837078,
        0.076614246,
        -0.030765701,
        0.06946893,
        -0.04846743,
        -0.014513135,
        0.013901048,
        0.036640476,
        0.07830911,
        -0.03368209,
        -0.053754743,
        -0.006383041,
        0.04265274,
        -0.022562794,
        0.054774076,
        0.021589909,
        -0.064304166,
        -0.61817133,
        -0.04916851,
        -0.019003954,
        -0.034899767,
        -0.15245229,
        -0.028624112,
        -0.035427317,
        0.013134341,
        0.06977043,
        -0.04066931,
        0.051123936,
        -0.02122012,
        0.0028224946,
        -0.046526913,
        0.04599472,
        0.01963678,
        0.015830435
      ]
    },
    {
      "id": "worldsfair-speaker-42",
      "conference": "worldsfair",
      "name": "Alexander Embiricos",
      "role": "Head of Enterprise Product",
      "company": "OpenAI",
      "twitter": "https://x.com/embirico",
      "talks": [
        {
          "title": "Opening Keynote: Topic TBD",
          "description": "TBD",
          "day": "Day 2 — Session Day 1",
          "time": "9:30am-9:50am",
          "type": "keynote"
        }
      ],
      "embedding": [
        -0.007136272,
        -0.0034173697,
        0.023807952,
        -0.0025967981,
        0.015995039,
        -0.061611548,
        -0.11879292,
        -0.011045486,
        0.011057355,
        -0.23357657,
        -0.014270082,
        -0.053402916,
        -0.008745852,
        -0.040838003,
        -0.06099736,
        0.010979618,
        0.043469332,
        -0.11764456,
        -0.021382643,
        -0.058052946,
        -0.073551886,
        -0.03217042,
        -0.04579237,
        0.012751681,
        -0.055616233,
        0.09022822,
        0.050946496,
        -0.008277734,
        -0.08946152,
        0.424133,
        -0.09835816,
        -0.00685871,
        -0.06269575,
        -0.002829865,
        0.060475416,
        -0.072922006,
        0.04475532,
        -0.05961113,
        -0.0035632406,
        0.057105582,
        0.0013442371,
        0.04415843,
        -0.059758626,
        0.072407186,
        0.0125169065,
        0.028205724,
        -0.057075974,
        -0.023725264,
        -0.0029724692,
        0.03736434,
        -0.016125716,
        -0.0019934396,
        -0.010029077,
        -0.02327395,
        0.1070734,
        0.0018019791,
        0.0071365754,
        0.035011694,
        0.10260177,
        -0.018714353,
        -0.10144782,
        -0.040040083,
        0.01819143,
        0.028564654,
        0.01757108,
        -0.046362188,
        0.065488525,
        -0.09018307,
        0.007851764,
        -0.038731515,
        0.03396619,
        0.055074826,
        -0.09099935,
        0.038668558,
        -0.10589539,
        -0.098663755,
        -0.01873718,
        0.04143074,
        -0.003664603,
        -0.025968153,
        0.052521743,
        0.058866195,
        -0.02860267,
        -0.029164553,
        -0.009980385,
        -0.089989886,
        -0.032512393,
        -0.016244138,
        -0.018392088,
        0.020725532,
        0.040656023,
        -0.015164762,
        -0.09958421,
        -0.00090124144,
        -0.056844447,
        0.041390985,
        -0.099708736,
        -0.0035049494,
        -0.0029179933,
        0.032148145,
        -0.014966712,
        -0.041758444,
        0.06148175,
        0.010889002,
        -0.085822634,
        -0.004903583,
        0.042127125,
        0.0032497952,
        0.032447454,
        0.005106325,
        -0.07978306,
        -0.637782,
        -0.016328176,
        0.016994251,
        -0.0064910487,
        -0.18738896,
        -0.03503492,
        -0.025640268,
        0.057694823,
        0.07824562,
        0.024816094,
        0.027776632,
        -0.09256737,
        0.082742184,
        0.045484375,
        0.07520499,
        -0.020475093,
        -0.021252973
      ]
    },
    {
      "id": "worldsfair-speaker-43",
      "conference": "worldsfair",
      "name": "Romain Huet",
      "role": "Head of Developer Experience",
      "company": "OpenAI",
      "twitter": "https://x.com/romainhuet",
      "talks": [
        {
          "title": "Opening Keynote: Topic TBD",
          "description": "TBD",
          "day": "Day 2 — Session Day 1",
          "time": "9:30am-9:50am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.0037197035,
        0.0250097,
        0.078777894,
        -0.02882792,
        0.09252266,
        -0.05685035,
        -0.0775886,
        0.0023196102,
        0.063661404,
        -0.2176033,
        0.0009366682,
        -0.04655681,
        -0.028001567,
        -0.025936017,
        -0.108866125,
        -0.009978181,
        0.02778286,
        -0.053237658,
        -0.024572948,
        -0.040220473,
        -0.091257766,
        -0.030704929,
        -0.012585118,
        0.07572326,
        -0.017299738,
        0.11289698,
        0.058958765,
        0.05075397,
        -0.14491883,
        0.42749545,
        -0.07178199,
        -0.025378637,
        -0.02523266,
        -0.008479881,
        0.047299404,
        -0.032439493,
        0.003815988,
        -0.023533981,
        -0.029662894,
        0.056519978,
        0.014667304,
        0.024522198,
        -0.0699958,
        0.07226485,
        -0.04779618,
        -0.027382305,
        -0.07871131,
        -0.04536883,
        -0.007581684,
        0.030838976,
        -0.067762665,
        -0.025712512,
        -0.013601215,
        -0.009050992,
        0.11734069,
        0.043058198,
        -0.0005479117,
        -0.0044010896,
        0.10325011,
        -0.0245841,
        -0.114365675,
        -0.045583412,
        0.0040973714,
        0.009589412,
        -0.034593903,
        -0.07021188,
        0.06367553,
        -0.06666184,
        -0.042384665,
        -0.025308335,
        0.043225415,
        0.042336766,
        -0.07806536,
        -0.027271084,
        -0.093452804,
        -0.012428567,
        -0.038296398,
        0.08443304,
        0.026566233,
        -0.066238806,
        -0.006077459,
        0.046386838,
        -0.054293487,
        -0.03537472,
        -0.00057705166,
        -0.102894396,
        -0.045799978,
        0.021393891,
        0.045622565,
        0.013560879,
        0.046861257,
        -0.09984294,
        -0.110097416,
        0.043248646,
        -0.097825326,
        0.025553325,
        -0.07132973,
        0.022676803,
        0.0063123605,
        -0.036145,
        -0.011010304,
        -0.038044766,
        -0.027257755,
        -0.021730796,
        0.008033237,
        -0.046682335,
        0.05994057,
        -0.01155231,
        0.039944954,
        -0.0021037606,
        -0.100623325,
        -0.6216779,
        0.037132483,
        0.043518662,
        -0.006462354,
        -0.16555598,
        -0.016400902,
        -0.008003899,
        0.06144272,
        0.097569965,
        -0.017138144,
        0.01607982,
        -0.022234883,
        0.02132906,
        0.028834654,
        0.09596789,
        -0.00023963161,
        -0.014135878
      ]
    },
    {
      "id": "worldsfair-speaker-44",
      "conference": "worldsfair",
      "name": "Thom Wolf",
      "role": "CTO",
      "company": "Hugging Face",
      "twitter": "https://x.com/Thom_Wolf",
      "talks": [
        {
          "title": "Thom Wolf keynote",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "9:50am-10:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        -0.016093576,
        0.00077946007,
        0.048044346,
        0.032711014,
        0.06172658,
        -0.01276031,
        -0.06347105,
        0.06269711,
        0.080197,
        -0.20876001,
        -0.06792161,
        -0.079815276,
        0.011245523,
        -0.045124106,
        -0.08826577,
        0.052864593,
        0.06438611,
        -0.060368985,
        -0.046843592,
        -0.055671796,
        -0.03429303,
        -0.026315602,
        -0.03308094,
        0.1262951,
        -0.05684438,
        0.1303131,
        0.01676125,
        0.06125897,
        -0.14010216,
        0.43814802,
        -0.07314729,
        0.015831415,
        -0.0077346875,
        -0.059348635,
        0.097601384,
        -0.053685065,
        0.0071567264,
        0.0037885436,
        0.018031841,
        0.07047158,
        -0.01016574,
        0.049686085,
        -0.018998556,
        0.020102652,
        0.0017142281,
        0.00035203918,
        -0.110744655,
        0.00038889993,
        -0.05558749,
        0.033218946,
        -0.086175226,
        0.03521464,
        0.034353454,
        -0.0034235178,
        0.09098425,
        -0.04755588,
        0.03033033,
        -0.101543464,
        0.15602157,
        -0.031022709,
        -0.08709629,
        -0.053924803,
        0.016385932,
        0.02336604,
        -0.013454499,
        -0.04845925,
        0.023938674,
        -0.048131164,
        -0.011168569,
        -0.031082032,
        -0.00849971,
        -0.009056257,
        -0.08077212,
        0.054473817,
        -0.13040332,
        -0.05556254,
        0.0033912733,
        0.060418207,
        0.01764369,
        -0.054208606,
        0.02440357,
        0.005432858,
        -0.063971706,
        -0.0927367,
        0.06450478,
        -0.036270563,
        -0.087847166,
        -0.005863909,
        0.025126955,
        0.050754678,
        0.0037757603,
        0.019578341,
        -0.03469323,
        0.027550917,
        -0.01510356,
        -0.0006550179,
        -0.08157081,
        0.0477902,
        0.02137382,
        0.01298194,
        -0.033043306,
        0.028095538,
        0.033399858,
        -0.10838375,
        -0.030327085,
        0.012793518,
        0.03135226,
        0.049132198,
        0.05208118,
        0.013268107,
        -0.055987604,
        -0.6095677,
        -0.008536103,
        0.009222149,
        -0.07888582,
        -0.14303324,
        -0.027020339,
        -0.011070411,
        0.05443746,
        0.048729446,
        -0.04088866,
        0.045829352,
        -0.01061948,
        0.048926882,
        0.034516554,
        0.063610345,
        0.020161238,
        -0.03184013
      ]
    },
    {
      "id": "worldsfair-speaker-45",
      "conference": "worldsfair",
      "name": "Zixuan Li",
      "role": "Head of Z.ai",
      "company": "Z.ai",
      "twitter": "",
      "talks": [
        {
          "title": "GLM-5.2: Frontier Intelligence, Open Weights.",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:10am-10:30am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.011858055,
        0.048801843,
        0.0030003185,
        -0.01316814,
        0.006095182,
        0.016228506,
        -0.027219636,
        -0.016365053,
        0.09050752,
        -0.21517655,
        -0.037137058,
        -0.039683666,
        -0.01665989,
        -0.016721446,
        -0.099235445,
        0.028352765,
        0.06399134,
        -0.03432063,
        -0.06368555,
        -0.0604925,
        -0.06455521,
        -0.040968485,
        -0.054332543,
        0.064763516,
        -0.04830052,
        0.03058867,
        0.011159763,
        -0.013228796,
        -0.10619932,
        0.43212816,
        -0.026211172,
        -0.020638539,
        0.009762818,
        0.033668853,
        0.061003834,
        -0.056554597,
        0.041543912,
        -0.078340605,
        0.036253575,
        0.07707272,
        0.05437748,
        0.020762341,
        -0.038372975,
        0.050615106,
        -0.012692785,
        -0.016742991,
        -0.07756306,
        0.012884935,
        -0.020722037,
        -0.00011864261,
        -0.055645704,
        -0.026097775,
        0.026217438,
        -0.018890463,
        -0.0023675463,
        0.03151221,
        -0.0026895178,
        -0.02333075,
        0.094991334,
        0.046055444,
        -0.15633352,
        0.02381168,
        0.05585913,
        0.11362242,
        0.016518785,
        -0.020930985,
        -0.018789455,
        -0.062441554,
        0.0006107366,
        -0.021478143,
        0.022301128,
        -0.015816305,
        0.0032171526,
        -0.0033433905,
        -0.12178229,
        -0.0720535,
        0.039283514,
        0.05441051,
        -0.01993641,
        -0.026024511,
        0.008343814,
        0.06505522,
        -0.029764779,
        -0.052741323,
        0.0058961827,
        -0.023189643,
        -0.06259562,
        0.073215745,
        -0.044788934,
        0.059967406,
        0.048720334,
        -0.027026488,
        -0.045172743,
        0.019870088,
        -0.150322,
        0.07715787,
        -0.053076256,
        0.060852807,
        0.045916725,
        0.015232871,
        -0.02691757,
        0.04004259,
        0.076651834,
        -0.010227974,
        -0.095531054,
        -0.041124925,
        0.014070444,
        -0.010252803,
        0.0126022035,
        -0.035852518,
        -0.103477515,
        -0.6521744,
        0.020259235,
        -0.025707768,
        -0.07121316,
        -0.15894875,
        0.018711885,
        0.0029829575,
        0.041422762,
        0.0723037,
        -0.044524454,
        0.022359781,
        -0.02251755,
        0.007638081,
        0.037208807,
        0.05093644,
        0.02720336,
        0.02836105
      ]
    },
    {
      "id": "worldsfair-speaker-46",
      "conference": "worldsfair",
      "name": "Jason Liu",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Codex Maxxing",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "Setting Yourself Up for Success — Part 1",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        },
        {
          "title": "Setting Yourself Up for Success — Part 2",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        },
        {
          "title": "Setting Yourself Up for Success — Part 3",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.004852337,
        0.10333629,
        -0.018414823,
        0.023447545,
        0.08523444,
        -0.021817556,
        -0.03892008,
        0.0063860253,
        0.05346739,
        -0.21090566,
        -0.07619197,
        -0.019456176,
        0.03719837,
        0.069623955,
        0.008376854,
        0.03935075,
        0.10938655,
        -0.12215153,
        -0.07243847,
        -0.09094053,
        -0.078353636,
        -0.14299828,
        -0.059040684,
        0.07429316,
        -0.07155254,
        0.113885544,
        0.017606864,
        0.015795197,
        -0.12295436,
        0.42625707,
        -0.038568992,
        0.021494662,
        -0.008710289,
        0.0202294,
        0.042583678,
        -0.050800037,
        0.0007021719,
        0.014462118,
        0.08266446,
        0.07938705,
        -0.015541972,
        -0.008349655,
        -0.017642012,
        0.051825322,
        -0.04384212,
        0.05562479,
        -0.104464345,
        -0.022492023,
        0.019996334,
        0.0014267822,
        -0.044093948,
        0.0011792572,
        0.024671683,
        -0.056117974,
        -0.053104773,
        0.020048792,
        -0.04980539,
        -0.030292915,
        0.090288915,
        0.028292429,
        -0.064530775,
        0.008866026,
        -0.05902938,
        0.009784805,
        0.007544148,
        -0.060268242,
        0.004899334,
        -0.038872506,
        0.008024595,
        -0.049370605,
        0.053240605,
        0.05150106,
        -0.059692357,
        0.0013198372,
        -0.07668642,
        -0.13647483,
        -0.00727054,
        0.09274217,
        -0.047996107,
        0.03860761,
        0.01343318,
        0.026802287,
        -0.05513099,
        -0.047535162,
        -0.011664164,
        0.02093791,
        -0.07273687,
        0.10577725,
        -0.08141769,
        0.04299088,
        0.0020208566,
        0.007728414,
        -0.022520022,
        -0.024991207,
        -0.061295856,
        0.059920844,
        -0.019740613,
        0.059188236,
        0.02571344,
        -0.020834576,
        -0.008167835,
        -0.027360164,
        0.041115746,
        0.002873997,
        0.002643812,
        -0.07650546,
        -0.02964586,
        -0.0570599,
        -0.060590703,
        0.03899386,
        -0.010733723,
        -0.6271278,
        -0.008409095,
        -0.00866188,
        -0.0014665129,
        -0.15332644,
        0.030034512,
        -0.012026939,
        -0.03589683,
        0.0790366,
        -0.017155625,
        0.03997141,
        0.05558726,
        -0.044048987,
        -0.058028154,
        -0.024187911,
        0.05883695,
        -0.013731929
      ]
    },
    {
      "id": "worldsfair-speaker-47",
      "conference": "worldsfair",
      "name": "Vinoth Govindarajan",
      "role": "Member of Technical Staff",
      "company": "Open AI",
      "twitter": "https://x.com/iamvinoth",
      "talks": [
        {
          "title": "Your Agent Didn’t Fail. Your Harness Did.",
          "description": "AI agents do not fail only because the model is wrong. Many production failures happen in the harness around the model: state is not persisted, two runs mutate the same session, a tool call never returns, an approval loses scope, or an internal success never becomes user-visible proof. This talk uses OpenClaw as a public case study to examine real harness failure modes and extract a reusable production model for AI engineers. We will look at how events enter an agent system, how session state is rehydrated, why single-writer lanes and throttles matter, and why tool execution needs scoped approvals and auditable receipts. The core idea is simple: a model proposes, the harness commits, and the receipt proves it. Attendees will leave with a practical 'run receipt' audit they can apply to their own agents: what woke it up, which state did it inherit, what authority did it use, what executed, and what evidence survived.",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.07540662,
        0.021523543,
        0.05352179,
        0.050548,
        0.024407012,
        -0.026354041,
        -0.08852964,
        0.018734518,
        0.027269086,
        -0.24283421,
        -0.021514026,
        -0.035054903,
        0.022573434,
        -0.034726955,
        0.0035942493,
        -0.04982046,
        0.10387474,
        -0.06898885,
        -0.08399938,
        -0.05975995,
        -0.042380154,
        -0.0917312,
        -0.020518437,
        0.049889892,
        -0.0469576,
        0.047596708,
        0.009544261,
        0.08507518,
        -0.07727468,
        0.43743622,
        -0.12805314,
        0.015905594,
        0.0033857115,
        -0.078473195,
        0.08431311,
        -0.110518865,
        0.0025365974,
        0.009387495,
        0.0039459304,
        0.0580887,
        -0.004665068,
        0.055191796,
        -0.037451815,
        0.046183173,
        0.0073179454,
        -0.02788298,
        -0.036082555,
        0.019947851,
        -0.018789735,
        -0.05876181,
        -0.08435523,
        -0.058115076,
        0.08822599,
        -0.0036886248,
        0.072854854,
        0.021212563,
        -0.011968981,
        -0.061695363,
        0.1011051,
        -0.013791037,
        -0.08176059,
        0.05620373,
        0.022836339,
        -0.06299259,
        0.022475647,
        -0.066453464,
        0.044626106,
        -0.08074791,
        -0.01379512,
        0.005676484,
        0.053154018,
        0.025006227,
        -0.06707702,
        -0.0025164625,
        -0.011756935,
        -0.14200792,
        -0.024222074,
        0.012618984,
        -0.008440286,
        -0.05952605,
        0.04302242,
        0.07115367,
        -0.011689139,
        0.020577488,
        0.005739583,
        -0.03551985,
        -0.049699504,
        0.02976726,
        -0.081524454,
        0.07649666,
        0.04561567,
        -0.03852192,
        -0.07698367,
        -0.017793184,
        -0.07474145,
        -0.00001991992,
        -0.043641336,
        0.08507259,
        -0.024930237,
        -0.009353182,
        -0.024502533,
        -0.0033549669,
        0.03372657,
        0.005357237,
        -0.04485098,
        -0.079860196,
        0.0015007285,
        0.004700616,
        0.03709674,
        0.028065326,
        -0.04063776,
        -0.6224108,
        0.00894372,
        0.003652262,
        -0.025045224,
        -0.1541044,
        0.045050666,
        0.0077151717,
        0.053934667,
        0.08041896,
        -0.041252058,
        0.10432825,
        -0.0054997653,
        0.018631449,
        -0.037747756,
        0.032014996,
        0.06108273,
        0.03135467
      ]
    },
    {
      "id": "worldsfair-speaker-48",
      "conference": "worldsfair",
      "name": "Edo Liberty",
      "role": "Founder & Chief Scientist",
      "company": "Pinecone",
      "twitter": "",
      "talks": [
        {
          "title": "Pinecone 2.0",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.04198244,
        0.021059498,
        -0.005264144,
        0.0552831,
        -0.022044742,
        -0.04714432,
        -0.1690533,
        0.03428109,
        0.008796764,
        -0.18657617,
        -0.058860376,
        -0.07103436,
        -0.013473461,
        0.047432147,
        -0.023781521,
        0.023236085,
        0.1392741,
        0.0038875386,
        -0.09709016,
        -0.078755744,
        -0.03489304,
        -0.07092423,
        -0.018351933,
        0.018045988,
        -0.039173644,
        0.052670624,
        0.03403766,
        0.0047620754,
        -0.1155385,
        0.5000215,
        -0.020791585,
        -0.06046151,
        0.035232574,
        -0.0377073,
        0.044089973,
        -0.048765294,
        -0.012209345,
        -0.023674162,
        -0.04782113,
        0.08193569,
        0.02797913,
        0.00977588,
        -0.011994739,
        -0.04033998,
        0.0018535508,
        -0.08201759,
        -0.09254357,
        0.056244526,
        -0.047808632,
        -0.0122597115,
        -0.052708127,
        0.028911302,
        0.047272116,
        -0.008634401,
        0.089522846,
        0.043078214,
        -0.035348307,
        -0.07797108,
        0.1015061,
        -0.0066496143,
        -0.1331813,
        -0.023175165,
        -0.03457724,
        0.056233093,
        0.030230133,
        -0.051077556,
        0.028299235,
        -0.07730202,
        -0.06413671,
        -0.038392246,
        -0.0068405475,
        0.030531636,
        -0.06528344,
        -0.05612367,
        -0.111065246,
        -0.07114584,
        -0.0037692082,
        0.046749152,
        0.0020264662,
        -0.05505683,
        0.03157276,
        0.05968948,
        -0.038445584,
        -0.050739683,
        0.06917394,
        -0.044912215,
        -0.07629742,
        0.0137604475,
        0.011149878,
        0.014844845,
        0.07643765,
        0.014960126,
        -0.0063051526,
        -0.022176031,
        -0.07356691,
        0.05810329,
        -0.06916396,
        0.052616224,
        -0.0047398913,
        0.004934774,
        -0.01731926,
        0.025595501,
        0.033919744,
        0.05200343,
        -0.050778836,
        -0.077431254,
        -0.016123844,
        0.03860219,
        0.10517487,
        -0.06266587,
        -0.084696785,
        -0.5770766,
        0.029432153,
        -0.048326734,
        -0.013118455,
        -0.14562531,
        -0.03557753,
        -0.010466595,
        0.0153277945,
        0.0083904695,
        -0.009760005,
        0.036144186,
        -0.016925486,
        0.04348557,
        0.038286056,
        0.07649119,
        -0.054842703,
        -0.043911595
      ]
    },
    {
      "id": "worldsfair-speaker-49",
      "conference": "worldsfair",
      "name": "Priyanka Phatak",
      "role": "Engineering Leader",
      "company": "Anthropic",
      "twitter": "https://x.com/PriyankaPhatak",
      "talks": [
        {
          "title": "Claude Managed Agents Workshop",
          "description": "Build an agent with Claude Managed Agents",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "Claude Managed Agents workshop",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        },
        {
          "title": "Claude Managed Agents workshop",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        },
        {
          "title": "Claude Managed Agents workshop",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06504423,
        0.05523213,
        0.051028382,
        0.0056430968,
        0.039899338,
        0.0052212398,
        -0.03239842,
        -0.012190026,
        0.040137704,
        -0.23605521,
        -0.015102651,
        -0.03890427,
        0.030032067,
        -0.03310771,
        -0.052983202,
        0.008067349,
        0.059298266,
        -0.023460956,
        -0.036257017,
        -0.11034108,
        -0.03629598,
        -0.08437388,
        0.014525209,
        0.07022581,
        -0.03928848,
        0.10678171,
        -0.020623881,
        0.029720517,
        -0.13897726,
        0.4661922,
        -0.13311058,
        -0.0022374194,
        0.079991356,
        -0.07861972,
        0.087428264,
        -0.055027124,
        -0.029510684,
        -0.048314832,
        -0.02218243,
        -0.024037052,
        0.0066202246,
        0.026894435,
        -0.037913524,
        0.012911922,
        -0.07884427,
        0.03101559,
        -0.10179071,
        0.04583725,
        -0.044684395,
        0.029421898,
        -0.05836862,
        -0.03142625,
        0.027143223,
        -0.010934373,
        0.026886826,
        0.012193715,
        -0.082779594,
        -0.10989343,
        0.11111635,
        -0.018703599,
        -0.026834622,
        0.014594126,
        -0.023092547,
        0.019916415,
        -0.02375909,
        -0.058154624,
        0.04980422,
        -0.042128153,
        -0.024449946,
        -0.047929022,
        0.019909762,
        0.02158085,
        -0.002348886,
        0.09834174,
        -0.044461656,
        -0.1081893,
        -0.05585282,
        0.058486145,
        -0.011326045,
        -0.023805011,
        0.013647357,
        0.040030118,
        -0.094364084,
        -0.08609376,
        0.04580798,
        -0.035234608,
        -0.06487187,
        0.007716631,
        -0.03870382,
        0.037593972,
        -0.003634991,
        -0.009947917,
        -0.05111764,
        0.03488089,
        -0.089181155,
        0.03794603,
        -0.027551243,
        0.03631749,
        0.022276558,
        0.020826822,
        0.034135774,
        -0.0292642,
        0.02409814,
        -0.00124205,
        -0.02981607,
        -0.013563696,
        -0.015610381,
        0.014771704,
        0.0672304,
        -0.0046807467,
        -0.0684243,
        -0.5952327,
        0.04745719,
        0.00053075,
        -0.07231853,
        -0.17672803,
        0.037479136,
        -0.06335724,
        0.056826938,
        0.1394612,
        -0.044061054,
        0.036431886,
        -0.05482991,
        -0.036574904,
        -0.03989009,
        -0.0012270706,
        -0.0030610226,
        0.018249642
      ]
    },
    {
      "id": "worldsfair-speaker-50",
      "conference": "worldsfair",
      "name": "Ezra Tanzer",
      "role": "Director, Product Management",
      "company": "Snyk",
      "twitter": "",
      "talks": [
        {
          "title": "Agentic Development Security",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.018266827,
        -0.020558534,
        0.015106776,
        -0.010428573,
        0.016767219,
        0.019228144,
        -0.050498765,
        -0.029098377,
        0.039895702,
        -0.233028,
        -0.07064768,
        -0.029787848,
        -0.0176677,
        -0.02764659,
        -0.06410806,
        0.0011622604,
        0.13886018,
        -0.04910868,
        -0.07410517,
        -0.09154244,
        0.0076944632,
        -0.109036975,
        0.012571989,
        0.090749815,
        -0.04454836,
        0.14456853,
        -0.020077255,
        0.085104726,
        -0.11086068,
        0.50181687,
        -0.036607113,
        0.010883493,
        0.008196872,
        0.031592596,
        -0.008435454,
        -0.087577656,
        -0.058507133,
        -0.0053806026,
        0.005691205,
        0.034086153,
        -0.028652327,
        -0.034395292,
        -0.063885294,
        0.04979541,
        -0.024001239,
        -0.0020727292,
        -0.108682536,
        -0.0033606556,
        -0.0004973966,
        -0.008169763,
        -0.04672301,
        0.025528815,
        0.054730795,
        -0.032642562,
        0.026775481,
        -0.017150408,
        0.025428386,
        -0.065181985,
        0.022911254,
        -0.02479246,
        -0.073447645,
        -0.020571506,
        0.023729865,
        0.042251248,
        0.015375033,
        -0.033283204,
        0.04737269,
        0.005610945,
        0.050787617,
        -0.061307766,
        0.073835626,
        0.05753136,
        -0.07395046,
        0.10855001,
        -0.1148975,
        -0.043250717,
        0.008509517,
        0.08415738,
        -0.01665907,
        -0.014952015,
        0.05224946,
        0.018289704,
        -0.12076383,
        -0.07133266,
        0.011131602,
        -0.07434559,
        0.02277573,
        -0.001399032,
        0.021559818,
        0.016883185,
        0.031035693,
        0.018068062,
        -0.019860664,
        0.06003655,
        -0.058592908,
        0.037154544,
        -0.093192294,
        0.078146204,
        -0.020067723,
        -0.030396162,
        -0.026195673,
        -0.0814429,
        0.044829506,
        0.032621667,
        -0.036546636,
        -0.044865724,
        -0.050290402,
        -0.009012661,
        0.107544325,
        0.016691655,
        -0.04726675,
        -0.57169443,
        0.021099485,
        -0.078423664,
        -0.012612885,
        -0.14259209,
        0.010480811,
        -0.005961687,
        0.017374791,
        0.06460288,
        -0.012205735,
        0.08427145,
        -0.06316219,
        0.03418453,
        0.00026402617,
        0.037335712,
        0.0402661,
        -0.01794394
      ]
    },
    {
      "id": "worldsfair-speaker-51",
      "conference": "worldsfair",
      "name": "Kwindla Kramer",
      "role": "",
      "company": "Daily",
      "twitter": "https://x.com/kwindla",
      "talks": [
        {
          "title": "The New Primitives: Building AI-Native Software",
          "description": "In the future, every piece of software with a human-facing surface will be built from new, LLM-centric primitives. We are just starting to invent these new primitives, including subagents, very long context, dynamic UI generation, and conversational voice input.",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "Voice is the universal interface",
          "description": "Language models give us the ability to create natural language, conversational, interfaces for computers. We are seeing a rapid shift among early adopters to using general language instead of traditional user interfaces for tasks like writing code and editing spreadsheets. Join the cofounders of Pipecat, Gradium, and Daily as we discuss the future of realtime voice and AI interfaces. Voice is the most efficient input mode for natural-language systems, and often the most efficient output mode, as well. But good voice interfaces require a very high degree of conversational facility, intelligence, task-specific reliability, and robustness to real-world realities like multiple speakers and background noise. There's a long history of voice interfaces in science fiction: Star Trek, Iron Man, Her. We'll use these depictions of computing possibilities as a jumping off point for talking about the ideal voice interface. How close are we to being able to build these interfaces with today's models, hardware, orchestration tooling, and UI libraries? What are the most promising research directions? What did the movies get wrong, now that we actually have experience building natural language, open-ended, voice systems?",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04770069,
        0.07791842,
        0.07340938,
        0.028148176,
        0.052406743,
        -0.04142607,
        -0.06873435,
        0.009216607,
        0.090243466,
        -0.24216221,
        -0.016259778,
        -0.059028164,
        -0.016080253,
        -0.05627239,
        -0.028542528,
        0.042839926,
        0.064428575,
        -0.056179933,
        -0.062773384,
        -0.109721765,
        -0.04123707,
        -0.08485445,
        -0.018885527,
        0.010667792,
        -0.0742937,
        0.058885355,
        0.0058742967,
        0.05540173,
        -0.050709706,
        0.41808718,
        -0.08498037,
        0.010585725,
        0.04111587,
        0.009180576,
        0.016056059,
        -0.039177783,
        -0.06463156,
        -0.03076484,
        0.091357104,
        0.08713164,
        0.051162146,
        0.0453076,
        -0.07587057,
        -0.029985057,
        -0.03653706,
        -0.043543298,
        -0.0517438,
        0.033734202,
        0.015925217,
        -0.012287404,
        -0.036249846,
        0.015710702,
        0.03173554,
        0.036444455,
        0.038650084,
        0.066726536,
        0.016004438,
        -0.1046013,
        0.06309129,
        0.0018285501,
        -0.074030295,
        0.038554702,
        -0.03706791,
        0.03381327,
        -0.0017286,
        -0.068235405,
        0.036459494,
        -0.051234074,
        -0.047224145,
        -0.062659174,
        -0.00050161016,
        0.027600473,
        -0.08104725,
        0.09321458,
        -0.0498555,
        -0.117856495,
        -0.027145267,
        0.0755465,
        -0.044085365,
        0.014539378,
        -0.04172649,
        0.06634414,
        -0.059145197,
        -0.0950628,
        0.008697256,
        -0.08723521,
        -0.04482536,
        -0.021963434,
        -0.0504086,
        0.024466785,
        0.06884136,
        -0.021151176,
        -0.06499718,
        0.00600344,
        0.014499752,
        -0.0023029281,
        -0.034982346,
        -0.012891665,
        -0.013879991,
        0.06579039,
        -0.0064315223,
        0.01426754,
        0.049886674,
        -0.0243234,
        -0.024343532,
        0.021504158,
        0.026875729,
        0.022205714,
        0.04232337,
        0.014675137,
        -0.013040717,
        -0.6594769,
        0.03766381,
        0.027468154,
        -0.040860467,
        -0.14019786,
        0.04213743,
        0.027715063,
        0.057719786,
        0.053768586,
        0.0141944345,
        0.045732267,
        0.016592009,
        -0.011812223,
        0.02179581,
        0.08166606,
        0.061736073,
        -0.016812699
      ]
    },
    {
      "id": "worldsfair-speaker-52",
      "conference": "worldsfair",
      "name": "Sean Cai",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "State of Data",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.0903037,
        0.09185449,
        -0.08738489,
        -0.0060715284,
        0.05222101,
        -0.05222003,
        -0.13773386,
        0.028219927,
        0.056542937,
        -0.25568014,
        -0.04687491,
        -0.019287549,
        0.004878502,
        0.06609183,
        -0.065337345,
        0.026468834,
        0.0521107,
        -0.095387585,
        -0.06991167,
        -0.044124607,
        0.009275465,
        -0.09430445,
        -0.04640256,
        0.0053741755,
        -0.0691133,
        0.04821409,
        0.011457171,
        0.00011124512,
        -0.115748726,
        0.4537091,
        0.0034403328,
        -0.026638241,
        -0.07925599,
        -0.01722839,
        0.03321486,
        -0.046712767,
        0.0074099824,
        -0.034648053,
        0.041934602,
        0.12337779,
        0.054791335,
        0.040230446,
        -0.024222484,
        0.056337185,
        -0.009590115,
        -0.04697633,
        -0.088538446,
        -0.08801695,
        -0.01984896,
        -0.03295232,
        -0.044714496,
        0.006267282,
        0.022506049,
        -0.03425582,
        -0.05455153,
        0.038329683,
        -0.022629026,
        -0.054205645,
        0.051529594,
        -0.07183973,
        -0.20573948,
        0.020062491,
        -0.041367568,
        -0.028036224,
        0.027975908,
        -0.020940691,
        0.030744126,
        -0.072380014,
        0.015114954,
        -0.016964626,
        -0.038116407,
        -0.00369877,
        -0.034081146,
        0.008547348,
        -0.13941102,
        -0.023929082,
        0.04374503,
        0.036121637,
        -0.04804875,
        0.016181486,
        0.023460224,
        0.007199816,
        -0.08405702,
        -0.047116783,
        -0.0019794363,
        -0.011196852,
        -0.09361081,
        0.03475771,
        -0.0628232,
        0.1185081,
        0.029713709,
        -0.02379345,
        -0.0059424303,
        0.005355409,
        -0.07616322,
        0.048940375,
        -0.075166486,
        0.03335507,
        -0.002522849,
        0.031137817,
        -0.0050202752,
        -0.0027716025,
        0.011563917,
        0.0014450904,
        -0.049780548,
        -0.026191194,
        0.033575576,
        -0.045405783,
        0.08301709,
        -0.061807923,
        -0.04459805,
        -0.5857036,
        0.07248317,
        -0.01667207,
        -0.03679933,
        -0.126905,
        0.0037201012,
        -0.03032455,
        0.03444752,
        0.049729764,
        -0.061705526,
        0.036291685,
        -0.035954177,
        -0.022417966,
        -0.0074572135,
        0.057108283,
        0.002687384,
        -0.023457365
      ]
    },
    {
      "id": "worldsfair-speaker-53",
      "conference": "worldsfair",
      "name": "Nilofer Rajpurkar",
      "role": "",
      "company": "Stripe",
      "twitter": "",
      "talks": [
        {
          "title": "Inside the AI economy: What Stripe’s data reveals",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0011575781,
        0.047583845,
        0.027116954,
        0.051773846,
        0.015447815,
        -0.009988421,
        -0.08681889,
        0.022613276,
        -0.014882933,
        -0.24084552,
        0.0015540331,
        -0.057583507,
        -0.021852475,
        -0.0008225887,
        -0.10327506,
        -0.018636605,
        0.074958414,
        -0.043712363,
        -0.05220093,
        -0.06571845,
        -0.049654327,
        -0.0528524,
        0.052684672,
        0.056773573,
        -0.079866506,
        0.07113795,
        0.03272134,
        -0.023987845,
        -0.12976328,
        0.48841092,
        -0.08552234,
        -0.039393373,
        0.0064535565,
        -0.03613639,
        0.01850366,
        -0.03791532,
        0.038624216,
        -0.02272742,
        -0.022411931,
        0.07963558,
        -0.009828905,
        0.031052042,
        -0.074388854,
        0.057308204,
        0.032998852,
        0.0033952866,
        -0.05639991,
        -0.054452177,
        0.010032079,
        0.008223904,
        -0.068194866,
        -0.0057907314,
        0.01995336,
        -0.023493182,
        0.07206025,
        0.031685386,
        -0.06797538,
        -0.07260709,
        0.118607104,
        -0.0551048,
        -0.12683685,
        0.01610248,
        0.023725444,
        0.007864438,
        0.081580244,
        -0.10290759,
        0.039143384,
        -0.03816873,
        0.065374985,
        -0.034920346,
        -0.028261472,
        -0.027096646,
        -0.045508344,
        0.058793835,
        -0.052021213,
        -0.08766518,
        0.008786579,
        0.07548526,
        0.020280857,
        0.018706396,
        0.08133717,
        0.014447487,
        -0.10245406,
        0.018038627,
        0.024759607,
        -0.021947931,
        -0.031037891,
        -0.01953435,
        -0.044717524,
        0.081935726,
        0.018884337,
        -0.03560829,
        0.034050286,
        0.012907991,
        -0.0006801223,
        0.034471195,
        -0.112697415,
        0.0875496,
        -0.0037406448,
        0.047087528,
        0.02539422,
        -0.05160625,
        0.058204945,
        -0.0062118983,
        -0.055487253,
        -0.008075521,
        -0.017086769,
        -0.04236025,
        0.04199881,
        -0.05326551,
        -0.04179056,
        -0.58589816,
        0.01744888,
        0.0046177395,
        -0.017670052,
        -0.15638487,
        0.020624077,
        -0.05994387,
        0.060885504,
        0.13379979,
        0.014845808,
        0.04587972,
        -0.030842988,
        -0.011800365,
        0.018030431,
        0.057792425,
        0.0026970443,
        0.0019159998
      ]
    },
    {
      "id": "worldsfair-speaker-54",
      "conference": "worldsfair",
      "name": "Mark Myshatyn",
      "role": "",
      "company": "Genesis Mission",
      "twitter": "",
      "talks": [
        {
          "title": "The Genesis Mission - Accelerating Science and National Security through AI",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.06119325,
        0.041366834,
        -0.054095488,
        0.13232763,
        -0.011814795,
        -0.02323591,
        -0.028531957,
        -0.062192015,
        -0.027772602,
        -0.22912388,
        -0.028938714,
        -0.055229083,
        0.011725379,
        -0.0005177851,
        -0.115030386,
        0.024358563,
        0.0492407,
        0.0034419778,
        -0.04801857,
        -0.064515546,
        0.027715612,
        -0.12758687,
        -0.07587913,
        0.031593483,
        -0.030398818,
        0.02737261,
        0.017020034,
        0.06772221,
        -0.09122535,
        0.4890447,
        -0.043238092,
        -0.027550215,
        0.007815001,
        -0.025237063,
        0.05632966,
        -0.03172151,
        -0.05611253,
        -0.030235607,
        -0.008774466,
        0.05524752,
        0.019888619,
        0.031498633,
        0.010209015,
        0.024785755,
        -0.023007767,
        0.05504704,
        -0.06896613,
        -0.04611909,
        0.027459051,
        -0.0037872926,
        -0.028837139,
        0.02073603,
        -0.006614559,
        -0.029836839,
        0.08789329,
        0.040126104,
        -0.0017863258,
        0.048559655,
        0.09236776,
        -0.07778446,
        -0.11681687,
        0.0322478,
        0.014183726,
        0.014963181,
        0.046561103,
        -0.053824443,
        0.044363163,
        -0.05791537,
        -0.020778224,
        0.037665132,
        -0.014983595,
        0.0014060169,
        -0.065973066,
        0.039634414,
        -0.051358253,
        0.0022447365,
        -0.0006128601,
        0.021126963,
        -0.086353235,
        -0.028261716,
        -0.001198957,
        -0.013846768,
        -0.04524102,
        -0.088252366,
        0.068481505,
        0.024662614,
        -0.013976155,
        0.0787234,
        -0.014613365,
        0.0015046858,
        0.028581886,
        -0.019346818,
        -0.0018713058,
        0.0001226473,
        -0.06787035,
        0.020373147,
        -0.07918945,
        0.02419778,
        0.0276011,
        -0.07545574,
        -0.08170818,
        -0.008609557,
        0.103251286,
        -0.012793251,
        -0.007886162,
        -0.03514534,
        0.044430204,
        0.012941819,
        0.052754246,
        -0.06680326,
        0.018813526,
        -0.62161666,
        -0.062543884,
        -0.05453262,
        -0.01563034,
        -0.16012987,
        -0.07054541,
        -0.004896317,
        0.08587536,
        0.039850175,
        -0.0145938415,
        0.003661891,
        -0.035207495,
        0.012147671,
        0.05486207,
        -0.0010121207,
        0.041228835,
        -0.0026603334
      ]
    },
    {
      "id": "worldsfair-speaker-55",
      "conference": "worldsfair",
      "name": "Tereza Tížková",
      "role": "",
      "company": "Factory",
      "twitter": "",
      "talks": [
        {
          "title": "Rise of the Software Factory",
          "description": "The Stanford HAI 2024 AI Index reports a 30x productivity gap between AI leaders and laggards. The differentiator is not company culture, prompting technique or model selection, but the infrastructure. Organizations capturing outsized value from AI agents have machine-readable codebases, deterministic internal APIs, CI/CD pipelines with agent-addressable hooks, and permission models granular enough to scope exactly what an agent can touch. I believe the “agents as employees” framing is most useful if you operationalize it. An employee has persistent identity, episodic and semantic memory, scoped permissions that don’t get renegotiated every task, an audit trail, and a defined escalation path when things go wrong. Persistent computer use (with a stable execution environment that survives across steps) was the real inflection point that is making this possible. Some interesting production problems remain under-explored. How do you give an agent persistent identity across pull requests? How do you recover from partial failure mid-task without discarding completed work? How do you enforce code ownership policies when the author is a model? How do you bound token spend when pipelines spin up sub-agents recursively? This talk defines agent readiness as a concrete infrastructure checklist: structured codebases, deterministic APIs, per-agent scoped credentials, atomic and idempotent operations, structured execution traces, and explicit thresholds for when the agent stops and a human takes over. It presents research results in practice, and what are the steps organizations need to take to be fully agent-ready.",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.08259086,
        0.025197223,
        0.056413297,
        -0.04214081,
        -0.0044140085,
        0.0012922812,
        -0.04272883,
        -0.018939966,
        0.04390804,
        -0.23809332,
        -0.076044604,
        0.018773038,
        -0.035403777,
        -0.029850403,
        -0.07343389,
        -0.03547182,
        0.0849684,
        -0.052466102,
        -0.051246457,
        -0.061900854,
        -0.000117122756,
        -0.051683486,
        -0.017739087,
        0.043630473,
        -0.017973876,
        0.17165163,
        0.051041637,
        0.011131127,
        -0.07870276,
        0.4487432,
        -0.07849697,
        0.076020434,
        0.045946717,
        -0.00907207,
        0.064233266,
        -0.0059902566,
        -0.088510975,
        -0.04515454,
        0.08037535,
        0.059541106,
        -0.02873712,
        0.03424486,
        -0.017190501,
        0.00753782,
        -0.0335214,
        -0.009436851,
        -0.07137088,
        0.03966174,
        -0.05161013,
        0.003192611,
        -0.048399165,
        0.024153557,
        0.036570333,
        -0.010519861,
        0.12026683,
        0.011462528,
        -0.0044662375,
        -0.069741204,
        0.1362972,
        -0.058086526,
        -0.09979597,
        -0.048062492,
        0.04093386,
        0.03154269,
        -0.017616075,
        -0.06742398,
        0.04633895,
        -0.14029878,
        0.013459081,
        -0.09388591,
        0.038799386,
        0.026828652,
        -0.03570845,
        0.008457264,
        -0.11962681,
        -0.081472374,
        -0.021822762,
        0.08046265,
        0.054253843,
        0.018584467,
        0.02188824,
        0.041404914,
        -0.11667357,
        -0.02609818,
        0.032283988,
        0.027264092,
        -0.09078572,
        0.009061789,
        -0.024847263,
        -0.0037753233,
        0.07540264,
        0.030224696,
        -0.050495487,
        0.054532085,
        -0.11355461,
        -0.018122308,
        -0.051026694,
        0.00015835371,
        -0.02564637,
        0.07247102,
        -0.04656931,
        -0.029170157,
        0.039491832,
        -0.036197565,
        0.007900551,
        0.009469359,
        -0.028759085,
        -0.00447759,
        0.04495407,
        -0.022862935,
        -0.024886334,
        -0.58231574,
        -0.020096427,
        -0.0049099796,
        0.008063608,
        -0.15584654,
        0.03664055,
        0.031569924,
        0.04949878,
        0.11123799,
        -0.02304815,
        0.023013601,
        -0.068558775,
        -0.041389562,
        0.063429505,
        0.066093266,
        0.01110758,
        -0.0015069724
      ]
    },
    {
      "id": "worldsfair-speaker-56",
      "conference": "worldsfair",
      "name": "Ryan Dahl",
      "role": "Creator of Node.js & Deno; Deno Sandbox",
      "company": "Deno",
      "twitter": "https://x.com/rough__sea",
      "talks": [
        {
          "title": "The OS runtime personal agents need",
          "description": "Why personal agents that run untrusted LLM code need a sandboxed OS/runtime model, not just a compute sandbox.",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.051693708,
        0.044904124,
        0.011549544,
        0.0007580606,
        0.04727873,
        -0.027058879,
        -0.0891928,
        -0.048793882,
        0.09045246,
        -0.23918149,
        -0.0188395,
        -0.07104173,
        -0.0036725623,
        -0.045293625,
        -0.06966687,
        0.024283987,
        0.063342266,
        -0.019116726,
        -0.047969006,
        -0.07538598,
        0.0010688587,
        -0.05738849,
        -0.038954865,
        0.040840067,
        -0.022874542,
        0.06870791,
        -0.02543792,
        0.07786468,
        -0.120873794,
        0.42155722,
        -0.089249164,
        -0.009116203,
        0.029188767,
        -0.034497835,
        0.081973456,
        -0.0811741,
        0.005282868,
        -0.0031840503,
        -0.023685696,
        0.021363929,
        0.019337326,
        0.020659788,
        -0.06071917,
        -0.022643812,
        -0.07765337,
        0.037909757,
        -0.04581489,
        -0.022569025,
        0.0032375986,
        0.029218262,
        0.002643161,
        0.04450705,
        0.10697325,
        0.006505408,
        0.04101975,
        -0.019011691,
        0.04113155,
        -0.110017166,
        0.049919795,
        -0.006689576,
        -0.1152791,
        -0.023892432,
        0.019812893,
        -0.015629234,
        -0.009730049,
        -0.04639662,
        0.055224787,
        -0.042308796,
        0.041011184,
        -0.04578708,
        0.040442917,
        0.014628917,
        -0.058342505,
        0.03812602,
        -0.08339404,
        -0.104842275,
        -0.034771796,
        0.00571505,
        -0.023520278,
        -0.015891114,
        -0.027528081,
        0.068485215,
        -0.07124967,
        -0.053353895,
        0.06502416,
        -0.063046254,
        -0.02809619,
        0.027211493,
        0.016097346,
        0.052115075,
        0.04623635,
        -0.04552811,
        -0.037902515,
        0.09584034,
        -0.027190411,
        0.044832237,
        -0.09651982,
        -0.025712699,
        -0.078388035,
        0.04834816,
        -0.012021619,
        0.015027412,
        0.013491364,
        -0.055362526,
        -0.007691368,
        0.008015484,
        0.014971099,
        -0.021075187,
        0.08530321,
        -0.032435317,
        -0.024082592,
        -0.6396875,
        -0.059719585,
        -0.06009008,
        -0.03319168,
        -0.15983243,
        0.047369063,
        0.025498804,
        0.08023766,
        0.08610161,
        -0.04487868,
        0.10542294,
        -0.013100018,
        -0.02753573,
        -0.016911885,
        0.078777045,
        0.019240111,
        -0.0044931164
      ]
    },
    {
      "id": "worldsfair-speaker-57",
      "conference": "worldsfair",
      "name": "Jerry Liu",
      "role": "Co-founder and CEO",
      "company": "LlamaIndex",
      "twitter": "https://x.com/jerryjliu0",
      "talks": [
        {
          "title": "HOLD — Jerry Liu / LlamaParse: Vision & OCR",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.0043234774,
        0.08278038,
        0.014701186,
        0.07191154,
        0.04801222,
        -0.013973325,
        -0.08680028,
        0.0092437975,
        0.061416626,
        -0.21186002,
        0.020070799,
        -0.07208415,
        -0.00087127497,
        0.07632116,
        -0.04945289,
        -0.013318053,
        0.04542117,
        -0.08965721,
        -0.06108381,
        -0.02642743,
        -0.040896706,
        -0.017272625,
        -0.020838283,
        0.038902592,
        -0.047918722,
        0.10467008,
        0.030935053,
        0.055469364,
        -0.11414293,
        0.456537,
        -0.06767854,
        -0.036127627,
        -0.02544006,
        0.043129068,
        0.044215936,
        -0.062814415,
        -0.02164026,
        -0.031642128,
        0.023071885,
        0.052832313,
        0.011310748,
        0.030620512,
        -0.016213467,
        0.085393295,
        0.009032708,
        -0.009761262,
        -0.039702445,
        0.039458606,
        0.04543811,
        0.040966846,
        -0.041968346,
        -0.0012908719,
        -0.022073068,
        0.027263273,
        -0.018612763,
        -0.01532743,
        -0.04932466,
        -0.007490841,
        0.093277745,
        0.008753776,
        -0.138255,
        -0.036176734,
        0.021029325,
        -0.017062638,
        0.015749471,
        -0.018480612,
        0.012099867,
        -0.02226357,
        0.023173755,
        -0.08332015,
        0.026529025,
        0.013799814,
        -0.054970577,
        0.00027351233,
        -0.049634352,
        -0.054306682,
        -0.026916083,
        0.0832686,
        0.026448846,
        0.0049581956,
        -0.005242513,
        0.07297908,
        -0.091059946,
        -0.07606038,
        0.0067115524,
        -0.038490407,
        -0.04838553,
        0.07593675,
        -0.014626314,
        0.08492992,
        0.016184032,
        -0.03965589,
        -0.06768179,
        -0.04143257,
        -0.08291842,
        0.04000028,
        -0.076863416,
        0.015270374,
        -0.017486434,
        0.034329187,
        -0.043361325,
        -0.008548497,
        0.031859983,
        -0.010674901,
        0.008694412,
        -0.05594392,
        0.07740494,
        -0.08357339,
        0.02162822,
        -0.03589394,
        -0.07250052,
        -0.648324,
        0.029856332,
        -0.018643059,
        -0.030700758,
        -0.17530902,
        -0.019931477,
        -0.03654882,
        0.011614555,
        0.04302165,
        0.02420673,
        0.09324009,
        0.019925455,
        0.039471027,
        -0.02063274,
        0.060675696,
        0.0247378,
        0.025186324
      ]
    },
    {
      "id": "worldsfair-speaker-58",
      "conference": "worldsfair",
      "name": "Jo Kristian Bergum",
      "role": "CEO & co-founder",
      "company": "Hornet.dev",
      "twitter": "https://x.com/jobergum",
      "talks": [
        {
          "title": "The unreasonable effectiveness of BM25 for agentic search",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.011972613,
        -0.026080316,
        0.015808193,
        0.031940084,
        -0.0018258566,
        -0.03921115,
        -0.06779648,
        0.08681063,
        0.061291758,
        -0.23490262,
        -0.06746345,
        -0.0072009573,
        0.0053679054,
        0.030213475,
        -0.12212665,
        0.0137939155,
        0.09980914,
        -0.11079847,
        -0.05661877,
        -0.03380712,
        -0.019329783,
        -0.061755516,
        -0.013713247,
        0.023962848,
        -0.03339145,
        0.110004276,
        -0.0087262625,
        0.04822904,
        -0.059067562,
        0.5142237,
        -0.05164515,
        0.019021217,
        0.047673095,
        -0.009978454,
        0.0129360035,
        -0.0904663,
        0.017439818,
        0.0045255264,
        0.018290736,
        0.09626559,
        -0.06313211,
        0.05791468,
        -0.018793339,
        0.025179876,
        -0.031209175,
        0.03547758,
        -0.060310263,
        0.028148351,
        0.02741815,
        0.014103549,
        -0.09880739,
        0.027090818,
        0.08293003,
        0.025030352,
        0.039709408,
        0.039485753,
        0.07843226,
        -0.06050507,
        0.080414295,
        0.05831029,
        -0.061236948,
        -0.054078262,
        -0.01395831,
        0.040443406,
        -0.010838048,
        -0.0016635569,
        0.018307706,
        -0.05415694,
        0.00036490013,
        -0.07424836,
        0.04273925,
        0.0117986305,
        -0.034032363,
        -0.013546636,
        -0.03605253,
        -0.08616586,
        0.059212983,
        0.07511017,
        -0.011168562,
        -0.015998073,
        -0.005640921,
        0.024307009,
        -0.03795905,
        -0.07215583,
        0.054398037,
        -0.03779574,
        -0.08879683,
        0.038445257,
        -0.03904461,
        0.10821649,
        0.04723874,
        -0.037143722,
        -0.02462342,
        0.032106172,
        -0.023012439,
        -0.011995722,
        -0.024894541,
        -0.0000070602555,
        0.008392729,
        -0.043171834,
        -0.0604751,
        0.0038943745,
        0.039475072,
        -0.02861917,
        -0.02173794,
        -0.028499348,
        0.051555682,
        -0.027807651,
        0.02686526,
        -0.022790415,
        -0.03139682,
        -0.6127387,
        -0.052284554,
        -0.022696309,
        0.00008591735,
        -0.15562679,
        0.017656649,
        0.04856423,
        0.027631514,
        0.036033206,
        -0.016683623,
        0.039011378,
        -0.0148986885,
        0.044704493,
        0.012936332,
        0.045232974,
        0.005272944,
        0.023254449
      ]
    },
    {
      "id": "worldsfair-speaker-59",
      "conference": "worldsfair",
      "name": "Ethan Cha",
      "role": "",
      "company": "Carlyle",
      "twitter": "",
      "talks": [
        {
          "title": "Dual-Surface Architecture: Serving Humans and Agents from the Same Tool Layer",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.046572007,
        0.00212601,
        -0.013075616,
        0.045451563,
        0.008180482,
        -0.011961941,
        -0.031134414,
        0.038267206,
        -0.055092957,
        -0.22449745,
        -0.03843245,
        -0.090939775,
        0.0076228515,
        -0.012593683,
        -0.03507189,
        0.06064228,
        0.045681335,
        -0.054276187,
        -0.13635886,
        -0.055578873,
        -0.059169047,
        -0.07768117,
        -0.0130764125,
        0.049796216,
        -0.035682086,
        0.14769316,
        -0.00973379,
        0.08749678,
        -0.066725336,
        0.45286813,
        -0.064096875,
        0.040569264,
        -0.060424954,
        0.004408883,
        0.021621197,
        -0.079517566,
        0.035061646,
        -0.006424037,
        0.087023064,
        0.032596227,
        -0.018587856,
        0.032919757,
        -0.031392094,
        0.042461373,
        -0.038140424,
        0.0032422692,
        -0.074649006,
        -0.021941256,
        -0.046122503,
        -0.04687514,
        -0.11598923,
        0.011974582,
        0.03056093,
        -0.069408424,
        0.008451756,
        0.0072731394,
        -0.02539902,
        -0.067055956,
        0.079232305,
        -0.0004492334,
        -0.11000259,
        -0.018743053,
        0.029456327,
        0.02494541,
        0.03333675,
        0.021912498,
        0.008734392,
        -0.07082039,
        0.056079436,
        -0.100092515,
        0.009348214,
        0.058300007,
        -0.049527414,
        -0.0026926089,
        -0.08340944,
        -0.04287595,
        -0.0212884,
        0.086130224,
        -0.006339297,
        -0.013894722,
        0.020858891,
        0.014939816,
        -0.091039516,
        -0.11998282,
        0.0075790854,
        0.0031218554,
        -0.043321487,
        0.058018222,
        0.014569133,
        0.02330105,
        0.03795295,
        0.051187214,
        -0.037224956,
        0.0021583755,
        -0.07723067,
        0.07651701,
        -0.09421013,
        -0.010284999,
        -0.05434365,
        0.0373511,
        -0.047861904,
        -0.09142803,
        0.024029648,
        0.039089777,
        -0.09204357,
        -0.020580893,
        0.012862681,
        -0.07646731,
        0.03334786,
        -0.025492152,
        -0.042596977,
        -0.61401975,
        -0.033626225,
        0.0011059459,
        -0.0035055215,
        -0.14062053,
        -0.029500986,
        0.0075622424,
        0.051183324,
        -0.012828246,
        -0.0957473,
        0.04452338,
        0.012947256,
        0.00028416206,
        -0.0007152937,
        0.005595754,
        0.07949036,
        0.07129813
      ]
    },
    {
      "id": "worldsfair-speaker-60",
      "conference": "worldsfair",
      "name": "Valeria Wu",
      "role": "Product Manager",
      "company": "Google DeepMind",
      "twitter": "",
      "talks": [
        {
          "title": "Speech-to-Speech Model Research at Google DeepMind",
          "description": "Most voice interfaces today are built as a 3-way cascade system (ASR/LLM/TTS). This session explores the shift toward native speech-to-speech models that process audio end to end, focusing on product and research challenges in building real-time voice agents with fluid turn-taking, low latency, and enterprise-grade intelligence.",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.049511768,
        0.022596275,
        -0.021848423,
        0.04974215,
        -0.0052952105,
        -0.023387868,
        -0.07809669,
        0.0021526024,
        0.10273268,
        -0.25197378,
        -0.021925483,
        -0.08152749,
        0.0018049936,
        -0.032354053,
        -0.047701344,
        0.044726368,
        0.04587876,
        -0.09839135,
        -0.025513941,
        -0.067124926,
        -0.003866409,
        -0.049977843,
        -0.00587111,
        0.049249306,
        -0.06724595,
        0.019384233,
        0.050799493,
        0.031045904,
        -0.085748196,
        0.4466743,
        -0.120911,
        0.0006419734,
        -0.055792365,
        0.021518186,
        0.052761268,
        -0.064719014,
        -0.013489952,
        -0.01106955,
        0.046852347,
        0.07258631,
        -0.042587228,
        0.0066950643,
        -0.05671492,
        0.01414972,
        -0.0106054135,
        -0.029557081,
        -0.03115207,
        0.04247513,
        -0.059110124,
        0.08518371,
        -0.031128604,
        0.027609076,
        0.010808631,
        0.0030847294,
        0.01821493,
        0.037681237,
        0.0076310174,
        0.02271086,
        0.10145316,
        -0.0041975603,
        -0.12090052,
        0.11322295,
        0.030544778,
        0.0120129315,
        0.041387003,
        -0.09625792,
        -0.035875004,
        -0.081765614,
        -0.09642005,
        -0.026144644,
        0.00849825,
        0.009812495,
        -0.050470367,
        0.013078255,
        -0.04606721,
        -0.15097743,
        -0.027938398,
        0.04098255,
        -0.017422771,
        0.001996451,
        0.02063951,
        0.09502344,
        -0.08390253,
        -0.05541438,
        0.031695373,
        -0.023789573,
        -0.0933485,
        0.034454007,
        -0.06045843,
        0.055241726,
        -0.021525184,
        -0.042416308,
        -0.051739067,
        0.009693865,
        -0.050376944,
        0.039925106,
        -0.06193802,
        0.023488395,
        -0.025531933,
        0.04993511,
        -0.057954613,
        0.02504971,
        0.013384972,
        0.009294184,
        0.0016906777,
        0.02918048,
        0.00875019,
        0.02309623,
        0.060076,
        0.0061864494,
        -0.10559789,
        -0.6148502,
        0.05556713,
        0.0017079625,
        0.035936967,
        -0.15201089,
        -0.007588187,
        -0.03870409,
        0.0718009,
        0.07677031,
        -0.012768952,
        0.059728973,
        -0.03601322,
        -0.018536843,
        0.024275864,
        0.067471676,
        0.064884655,
        0.025181005
      ]
    },
    {
      "id": "worldsfair-speaker-61",
      "conference": "worldsfair",
      "name": "Jesse Lumarie",
      "role": "Software Engineer",
      "company": "Figma",
      "twitter": "https://x.com/jesselumarie",
      "talks": [
        {
          "title": "Building the engine while flying the plane — launching the Figma MCP server",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0055830395,
        -0.020676516,
        -0.0011437591,
        -0.023426834,
        0.049784824,
        -0.0037180996,
        -0.08528033,
        -0.06700869,
        0.07636207,
        -0.20372155,
        -0.055610612,
        -0.024242347,
        0.059999958,
        0.02768713,
        -0.07234911,
        0.05538532,
        0.09500147,
        -0.10631438,
        0.0004023677,
        -0.07477517,
        -0.03368852,
        -0.056829453,
        -0.036414713,
        0.07886848,
        -0.06900258,
        0.056091562,
        0.013693247,
        -0.01225036,
        -0.13967957,
        0.4860111,
        -0.08835416,
        0.0572323,
        0.056945182,
        -0.047911175,
        0.07468409,
        -0.04243817,
        -0.046915058,
        0.018250927,
        0.015573898,
        0.047873408,
        -0.09086101,
        0.031390928,
        -0.07510401,
        0.052694842,
        0.0003403843,
        0.029900257,
        -0.05197594,
        0.0013974954,
        0.0036565042,
        0.025397783,
        -0.09990962,
        -0.024989365,
        0.05081192,
        -0.0049315128,
        0.018299147,
        -0.007328089,
        0.009872335,
        -0.04606782,
        0.058515143,
        0.044234987,
        -0.058465376,
        0.033989627,
        0.035793442,
        0.016564595,
        0.046742946,
        -0.040801793,
        0.060522377,
        0.0043647294,
        0.018049031,
        -0.0077273003,
        0.041584514,
        0.0016728825,
        -0.04465866,
        0.032568704,
        -0.109019086,
        -0.057931207,
        -0.010254255,
        0.0341128,
        0.024356654,
        0.010314216,
        0.03820586,
        0.04107658,
        -0.063007265,
        -0.01580806,
        0.052865505,
        -0.047145303,
        -0.03923415,
        0.0740545,
        -0.027524926,
        0.036292564,
        -0.005560871,
        0.001684337,
        -0.013168121,
        0.03563053,
        -0.020232135,
        -0.0008908067,
        -0.100831926,
        0.030326158,
        0.008003559,
        0.0585375,
        -0.07565534,
        -0.02492311,
        -0.036462046,
        0.025099317,
        -0.005058613,
        -0.008565549,
        -0.023480225,
        -0.02043608,
        0.07370751,
        0.021967856,
        0.0043547694,
        -0.6445882,
        -0.023658033,
        -0.030666415,
        0.0033692473,
        -0.15329477,
        -0.015935717,
        -0.017777184,
        -0.026283383,
        0.051565256,
        0.004837724,
        0.039008003,
        -0.054055642,
        -0.0614135,
        -0.013809491,
        0.05528923,
        -0.013684702,
        -0.04102827
      ]
    },
    {
      "id": "worldsfair-speaker-62",
      "conference": "worldsfair",
      "name": "Ameya Bhatawdekar",
      "role": "",
      "company": "Braintrust",
      "twitter": "",
      "talks": [
        {
          "title": "Your Agent Evolved. Your Evals Didn't.",
          "description": "Knowing which generation your agent is in, which failure modes your current evals are blind to, and what to build next is the difference between shipping with confidence and flying blind. Agent architectures have evolved through six generations; prompt, chain, ReAct loop, workflow graph, modern agent loop, AI harness. And each one quietly breaks the eval strategy of the generation before it. A prompt-quality rubric won't catch a bad tool call; a trace scorer won't catch memory poisoning. Using a single SRE incident response agent threaded through every generation, this talk shows exactly where each architecture outgrows its evals and what you need to close the gap.",
          "day": "Day 2 — Session Day 1",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.050251957,
        0.078942694,
        0.042638525,
        0.042068113,
        0.013935778,
        -0.03396586,
        -0.090939835,
        -0.006425031,
        -0.02260871,
        -0.2148867,
        -0.03628349,
        -0.029665586,
        0.07879372,
        0.0024313738,
        -0.07303782,
        0.0016241921,
        0.119325496,
        -0.018753426,
        -0.106356286,
        -0.07344264,
        -0.04458755,
        -0.052245863,
        -0.024835037,
        0.020372538,
        -0.036762036,
        0.09988065,
        -0.016953148,
        0.100765616,
        -0.11503655,
        0.47115478,
        -0.1277364,
        -0.05545032,
        0.05266971,
        -0.040267676,
        -0.014381673,
        -0.09106023,
        0.01581318,
        0.011463038,
        -0.010047617,
        0.05897868,
        -0.0022346454,
        0.05508393,
        0.0081856195,
        0.05390915,
        -0.043225534,
        -0.009846924,
        -0.05642382,
        -0.023297293,
        -0.0013019632,
        -0.06075028,
        -0.051150393,
        0.0032734105,
        0.060701672,
        -0.050317112,
        0.012358205,
        0.0838763,
        0.013753524,
        -0.030356675,
        0.1262779,
        0.0030019772,
        -0.05578603,
        0.014134201,
        0.021945504,
        -0.029136896,
        0.029458908,
        -0.06851806,
        0.11429857,
        -0.1315902,
        -0.015031329,
        -0.02045599,
        -0.0088396445,
        0.0284065,
        -0.006031064,
        0.039072324,
        -0.041363593,
        -0.14132331,
        -0.04435331,
        0.10859523,
        -0.023463663,
        0.020383373,
        -0.009838804,
        0.03003428,
        -0.06047496,
        -0.07665532,
        0.027007781,
        0.026216844,
        -0.04623799,
        0.02000886,
        -0.04139538,
        0.04504401,
        0.012375069,
        -0.024543034,
        -0.05331151,
        0.005496568,
        -0.035758395,
        0.024165887,
        -0.07249444,
        0.039749473,
        -0.028699217,
        0.023020338,
        0.02669256,
        -0.018896133,
        0.090039,
        -0.0039136857,
        0.059035093,
        -0.04448713,
        0.040352672,
        0.045513563,
        0.022486247,
        0.017109435,
        0.025758145,
        -0.58954513,
        0.0032880665,
        -0.009984153,
        -0.03712771,
        -0.15785122,
        0.042782776,
        0.045185693,
        0.036315456,
        0.095062874,
        -0.04584791,
        0.057920557,
        -0.041864086,
        0.004930107,
        -0.031077849,
        0.06451812,
        0.07965456,
        0.008213389
      ]
    },
    {
      "id": "worldsfair-speaker-63",
      "conference": "worldsfair",
      "name": "Charlie Holtz",
      "role": "Founder & CEO",
      "company": "Conductor",
      "twitter": "https://x.com/charlieholtz",
      "talks": [
        {
          "title": "Everything is Conductor",
          "description": "Everything is Conductor now! I want to tell the story of how we came up with the original interface, what I think everyone (including us) is getting wrong and what's coming next.",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0031058304,
        0.04990185,
        0.048296582,
        -0.05133398,
        0.07668448,
        -0.037274834,
        -0.105465464,
        0.036851488,
        0.071572885,
        -0.21075997,
        -0.027015245,
        -0.06672346,
        -0.026560813,
        0.0197463,
        0.0084392615,
        0.0587188,
        0.06255216,
        -0.06770203,
        -0.10278307,
        -0.053158447,
        -0.047024593,
        -0.044285785,
        -0.11053146,
        0.036997553,
        -0.060595654,
        0.16669858,
        0.01712391,
        0.034988556,
        -0.09229059,
        0.46480337,
        -0.025136072,
        -0.04924892,
        -0.022568148,
        -0.022682779,
        0.06219511,
        -0.061701875,
        -0.009805502,
        0.023133257,
        0.009890038,
        0.06252104,
        -0.045782287,
        0.104023516,
        -0.032185834,
        -0.027158266,
        -0.012528216,
        -0.08782125,
        -0.119508654,
        -0.053443972,
        -0.007927981,
        0.07374365,
        -0.05371766,
        0.002326846,
        -0.017225927,
        0.033358917,
        -0.000717178,
        -0.06965321,
        -0.07026493,
        -0.001946674,
        0.07478161,
        -0.03536165,
        -0.053037584,
        -0.023439467,
        -0.0126577085,
        -0.020024855,
        -0.018714467,
        -0.0038704055,
        0.05561752,
        -0.021285562,
        0.031444576,
        -0.0880994,
        0.04990847,
        0.049217727,
        -0.05677952,
        -0.022306819,
        -0.09351001,
        -0.07519288,
        -0.02249857,
        -0.02896948,
        -0.011855931,
        0.025928833,
        0.010972913,
        0.023970468,
        -0.055881664,
        -0.07364541,
        -0.026266439,
        -0.071057744,
        -0.050491408,
        0.04758665,
        0.024195882,
        0.0876928,
        0.04240381,
        -0.027189298,
        0.042587318,
        0.006709945,
        -0.04574202,
        -0.015483312,
        -0.10743208,
        0.011607373,
        0.030463396,
        0.047768295,
        -0.056384586,
        0.026788265,
        -0.051675364,
        0.02534886,
        -0.06468716,
        -0.1419054,
        0.047154125,
        0.03305688,
        0.007218624,
        0.020845655,
        0.043750945,
        -0.599765,
        -0.01049954,
        0.04398419,
        0.010908221,
        -0.17160787,
        0.018858992,
        0.037867256,
        -0.0138890445,
        0.045742266,
        -0.004707965,
        0.03030544,
        -0.023403767,
        0.035415,
        0.009163808,
        0.05999913,
        0.04735187,
        0.030759396
      ]
    },
    {
      "id": "worldsfair-speaker-64",
      "conference": "worldsfair",
      "name": "Kenton Varda",
      "role": "Principal Engineer",
      "company": "Cloudflare",
      "twitter": "https://x.com/KentonVarda",
      "talks": [
        {
          "title": "Gadgets: Personal app vibe coding that is actually safe",
          "description": "We are entering the end game of Kenton's 15-year master plan. The architect of Cloudflare Workers, Durable Objects, Cap'n Proto, and Sandstorm.io, and the guy who coined the term \"Code Mode\", will demo Gadgets, an AI productivity suite which ties all these ideas together. We've all heard that the future is micro-apps customized for every niche, but how do we actually make that usable, how do we make it scale, and most importantly, how do we make it safe for even non-developers to use? Kenton will show how Gadgets solves these problems, including a sandbox design that makes it essentially impossible for apps to have vulnerabilities at all. He'll then open source it for your slop-forking pleasure.",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.048213936,
        0.097641125,
        0.031535007,
        0.012804382,
        0.017984161,
        0.005006973,
        -0.10895639,
        -0.036578607,
        0.069839746,
        -0.1930735,
        -0.07435645,
        -0.03775483,
        0.08233768,
        -0.056363806,
        0.022956433,
        0.025032638,
        0.05374673,
        -0.078927256,
        -0.020658573,
        -0.14072523,
        0.02437536,
        -0.1231183,
        -0.048181374,
        0.08270305,
        -0.030969795,
        0.076554075,
        0.0020707347,
        0.039894734,
        -0.039056756,
        0.43821523,
        -0.04854401,
        -0.0043091318,
        0.02331005,
        -0.036014352,
        0.04833581,
        -0.029312216,
        -0.0074158385,
        -0.0006951049,
        0.03556703,
        0.048427824,
        0.024213092,
        0.0027747862,
        -0.06688577,
        -0.040816996,
        -0.06660381,
        -0.020581812,
        -0.03435259,
        0.043066368,
        0.038717814,
        -0.000965769,
        -0.044380568,
        0.042593725,
        0.13565734,
        0.03497369,
        0.04153465,
        -0.04648935,
        0.004898476,
        -0.06751011,
        0.068036035,
        0.01147973,
        -0.054986075,
        0.011096458,
        -0.043397382,
        0.007312188,
        0.007609,
        -0.050370403,
        0.04602333,
        0.029847039,
        -0.028393934,
        -0.0277823,
        -0.0024056432,
        0.049028903,
        -0.0238269,
        -0.0061494177,
        -0.04372628,
        -0.12490866,
        -0.035538595,
        0.0024113038,
        -0.059870448,
        0.0014361334,
        0.024207795,
        0.03369887,
        -0.064587444,
        -0.009362938,
        0.0058604376,
        -0.023097927,
        -0.07975662,
        0.047837023,
        -0.001869786,
        0.082461126,
        0.003257586,
        -0.034859464,
        -0.055154312,
        0.04597981,
        -0.055841785,
        0.039564524,
        -0.01366687,
        0.0129448045,
        -0.03159003,
        0.03511475,
        -0.038109753,
        -0.02687489,
        0.03524219,
        -0.011760912,
        -0.059207674,
        0.017818224,
        -0.007897113,
        -0.00977713,
        0.04708809,
        -0.03989202,
        -0.027484989,
        -0.65679866,
        0.023941634,
        0.032484584,
        0.05330069,
        -0.13034283,
        0.045363665,
        -0.018474318,
        0.12691255,
        0.119598135,
        -0.046687555,
        0.06838727,
        -0.03897841,
        -0.012814185,
        -0.01605294,
        0.09704624,
        0.061827898,
        -0.0057327934
      ]
    },
    {
      "id": "worldsfair-speaker-65",
      "conference": "worldsfair",
      "name": "Merve Noyan",
      "role": "Developer Advocate",
      "company": "Hugging Face",
      "twitter": "https://x.com/mervenoyann",
      "talks": [
        {
          "title": "Skill issue: stop deploying vision language models, use them with Skills to build e2e vision apps on edge",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.017754443,
        0.0063375006,
        0.024897713,
        0.002667214,
        0.057149798,
        -0.011773319,
        -0.002336528,
        0.0022050913,
        0.06460302,
        -0.18081366,
        0.035169955,
        -0.025742978,
        0.024880707,
        0.011394316,
        -0.0885255,
        0.04222126,
        0.07716246,
        -0.09340005,
        -0.059949193,
        -0.1131595,
        -0.03948534,
        -0.035865195,
        0.009000807,
        0.05885843,
        -0.012552187,
        0.07936304,
        -0.022121156,
        0.012525148,
        -0.14158633,
        0.5055917,
        -0.11761587,
        -0.021961328,
        0.03940391,
        -0.060407966,
        0.02868754,
        -0.049410004,
        0.006928969,
        -0.0026147268,
        0.04011238,
        0.0780934,
        -0.0033271974,
        0.06014604,
        -0.037254404,
        0.07051759,
        -0.06890623,
        0.0097253295,
        -0.04010063,
        0.08485003,
        0.06583658,
        -0.0041720667,
        -0.08728462,
        -0.046649642,
        0.07057459,
        0.021834731,
        0.09665696,
        0.017583374,
        0.02523006,
        -0.081807345,
        0.08929488,
        -0.01316024,
        -0.060193628,
        -0.037747756,
        -0.014594926,
        0.0199511,
        -0.017309435,
        -0.0692018,
        0.08346453,
        -0.12554084,
        0.014837593,
        -0.023464555,
        -0.04340569,
        -0.021692239,
        -0.01982377,
        0.056900587,
        -0.023993202,
        -0.04752632,
        0.03079324,
        0.055574592,
        -0.022702165,
        -0.05249169,
        -0.016851338,
        0.05778204,
        -0.065602496,
        -0.064747594,
        0.042577703,
        -0.061641667,
        -0.05109971,
        0.05691301,
        -0.026285768,
        0.046706885,
        0.018922947,
        -0.0465952,
        -0.071311824,
        0.02662553,
        -0.09605441,
        0.043058548,
        -0.031227054,
        0.088737935,
        0.011397371,
        0.029657248,
        -0.05376701,
        -0.047452293,
        0.002432355,
        -0.029789103,
        -0.006433111,
        -0.031033088,
        -0.012746919,
        -0.013754087,
        0.04687217,
        -0.064821474,
        -0.044178516,
        -0.605479,
        0.0040389188,
        -0.047719985,
        -0.03300058,
        -0.15832807,
        -0.052709483,
        -0.008484964,
        0.019519849,
        0.034069076,
        -0.020726725,
        0.047399275,
        -0.046271738,
        0.017180827,
        -0.011594805,
        0.022471502,
        0.015005116,
        -0.0075050564
      ]
    },
    {
      "id": "worldsfair-speaker-66",
      "conference": "worldsfair",
      "name": "Will Bryk",
      "role": "Co-founder and CEO",
      "company": "Exa",
      "twitter": "https://x.com/WilliamBryk",
      "talks": [
        {
          "title": "The Search Engine for the Agentic Web",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.03949396,
        0.047964968,
        0.018240297,
        0.031882066,
        0.051266123,
        -0.0012138834,
        -0.11745533,
        -0.00031618858,
        -0.021048754,
        -0.20658877,
        -0.0406372,
        -0.06823148,
        0.03381859,
        0.00092836726,
        -0.057718694,
        0.052843496,
        0.073182724,
        -0.04833931,
        -0.045845944,
        -0.09833669,
        -0.07036797,
        -0.07977504,
        -0.04595734,
        0.0043865,
        -0.07109087,
        0.07740256,
        -0.023044916,
        0.04214546,
        -0.13868241,
        0.47798994,
        -0.0708139,
        -0.021971693,
        0.0031835258,
        0.061270844,
        -0.017242372,
        -0.057506267,
        0.026935568,
        -0.005658303,
        0.07553939,
        0.050318222,
        -0.0015866032,
        0.07044119,
        -0.0409414,
        -0.005532415,
        -0.0022605977,
        0.030905688,
        -0.059308745,
        -0.0077394196,
        -0.025460368,
        0.005737684,
        -0.08456716,
        0.06330223,
        0.048361745,
        -0.018045582,
        0.06863283,
        -0.0072906464,
        -0.075439066,
        -0.08616064,
        -0.010026766,
        -0.026621116,
        -0.06758681,
        0.03419052,
        0.0072487066,
        0.04728312,
        0.046036486,
        -0.09976564,
        0.03676437,
        -0.04485565,
        -0.021608999,
        -0.063278824,
        -0.057209272,
        -0.012118326,
        -0.012050374,
        0.046619438,
        -0.04221559,
        -0.081261136,
        -0.0043824385,
        0.051428378,
        0.021154506,
        -0.060419932,
        -0.012087969,
        0.018973418,
        -0.03921959,
        -0.057002973,
        -0.023626236,
        -0.05163604,
        -0.008791194,
        0.027986493,
        -0.057213146,
        0.056008562,
        0.01116829,
        0.009181793,
        -0.07548793,
        -0.051006116,
        -0.043207247,
        0.0074098874,
        -0.055237494,
        0.07355573,
        0.04106737,
        -0.0045145126,
        -0.024285605,
        0.014595945,
        0.06768198,
        0.004776566,
        -0.047178783,
        -0.10153564,
        0.0666198,
        0.0070455945,
        0.0623731,
        -0.08101418,
        -0.08930633,
        -0.61636376,
        -0.04397938,
        -0.030798072,
        -0.04931212,
        -0.16546085,
        0.013294101,
        0.0652183,
        0.018830044,
        0.018906077,
        -0.05934334,
        0.0745555,
        0.0030903379,
        0.045542125,
        0.016062615,
        0.077548794,
        0.006958223,
        -0.012081659
      ]
    },
    {
      "id": "worldsfair-speaker-67",
      "conference": "worldsfair",
      "name": "Lovina Dmello",
      "role": "Senior Software Developer",
      "company": "NVIDIA",
      "twitter": "",
      "talks": [
        {
          "title": "Your LLM Stack Is a 2008 Database With Better Marketing: Why ML Security Is Dominated by Misconfiguration, Not Missing Features",
          "description": "ShadowRay exposed over a billion dollars of data through a missing authentication check. It wasn't a zero-day. It wasn't a clever new attack class. It was a default config someone never flipped off. That story is not the exception in production ML, it's the rule. We synthesized 139 peer-reviewed papers on production ML security across access control, runtime security, infrastructure, and operations. Five findings stood out, and one of them upends how most teams think about ML security: - Misconfiguration, not missing features, is the dominant failure mode. The mechanisms exist. Teams aren't using them, or are using them wrong. - Adversarial defenses impose 15–30% inference overhead, which is why almost no production system actually runs them. - ML-specific security tooling lags general DevOps tooling by years. - Security, data-science, and ops teams operate in expertise silos that create persistent gaps no single team can see. - LLM and multi-tenant GPU threats are evolving faster than defenses (prompt injection, RAG poisoning, GPU side channels). This talk walks through the four-pillar defense-in-depth framework, the six-category threat taxonomy that maps each attack to its primary and secondary defenses, and a four-level security maturity model that matches overhead budgets to deployment contexts. You leave knowing where your stack actually sits and which 3 misconfigurations account for most of the risk.",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.02643861,
        0.058114715,
        0.024412667,
        0.037136916,
        0.051382978,
        -0.0448888,
        -0.049304616,
        -0.00049464876,
        0.072451964,
        -0.20159002,
        -0.029705763,
        -0.061092306,
        -0.025932929,
        -0.047983516,
        -0.05840942,
        0.02570512,
        0.07444196,
        -0.042045955,
        -0.061078772,
        -0.1129012,
        -0.025556253,
        -0.06899949,
        0.018203435,
        0.1087,
        -0.13148643,
        0.05282397,
        0.021290965,
        0.07403159,
        -0.10034739,
        0.47293475,
        -0.10093905,
        0.03618143,
        0.04581945,
        -0.065230906,
        0.075722516,
        -0.08663211,
        -0.024684357,
        0.0055619413,
        0.008758042,
        0.062215786,
        0.04366428,
        0.050089322,
        -0.089002505,
        0.05429266,
        -0.011326373,
        -0.047244284,
        -0.051919527,
        0.045785755,
        0.020979742,
        -0.02357423,
        -0.0591323,
        -0.073691055,
        0.11089585,
        -0.015096588,
        0.021080172,
        0.022317791,
        0.014919404,
        0.004342011,
        0.08939327,
        -0.06400142,
        -0.10068917,
        -0.016584922,
        0.028273903,
        0.056047678,
        0.02154748,
        -0.095405534,
        0.05218289,
        -0.057302255,
        -0.0027663205,
        0.012452292,
        0.043093026,
        0.03231394,
        -0.03511027,
        0.06361398,
        -0.036121175,
        -0.11668765,
        -0.052721057,
        0.05979285,
        -0.017739816,
        0.017916884,
        0.05147736,
        0.046090458,
        -0.043084867,
        -0.05801686,
        0.049062416,
        -0.022479655,
        -0.1008863,
        0.009630196,
        -0.058759626,
        0.08743791,
        0.014036366,
        -0.036813058,
        -0.011684479,
        0.042636923,
        -0.06885662,
        0.020615948,
        -0.05214428,
        0.11923891,
        -0.032438345,
        0.010317916,
        0.06784939,
        -0.008522192,
        0.055734772,
        0.002128976,
        -0.0020395564,
        -0.043128565,
        -0.054551344,
        -0.029684685,
        0.056474455,
        0.008523644,
        -0.015114068,
        -0.5921457,
        -0.020157048,
        -0.00802898,
        -0.074432746,
        -0.13623548,
        0.04578951,
        -0.042541813,
        0.052619457,
        0.033158924,
        -0.010083261,
        0.05843757,
        -0.06582558,
        -0.04175459,
        0.004685123,
        -0.023895307,
        0.051222444,
        -0.018964171
      ]
    },
    {
      "id": "worldsfair-speaker-68",
      "conference": "worldsfair",
      "name": "Deniz Birlikci",
      "role": "",
      "company": "NeoLabs",
      "twitter": "",
      "talks": [
        {
          "title": "FrontierCode needs Frontier Data Quality",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09084122,
        -0.014489337,
        0.035219166,
        0.007929338,
        -0.018816236,
        0.03217941,
        -0.08438891,
        0.0296768,
        0.09406197,
        -0.21954621,
        -0.010684867,
        -0.012912652,
        -0.011031157,
        0.06979993,
        -0.05282318,
        -0.02419306,
        0.12736191,
        -0.042739216,
        -0.033123516,
        -0.087796375,
        -0.08324952,
        -0.04802972,
        -0.033487573,
        -0.029082742,
        -0.028255802,
        0.067494005,
        0.013238389,
        0.009715071,
        -0.07727943,
        0.54358435,
        -0.056945514,
        0.016130853,
        0.05581271,
        -0.031126458,
        0.017398836,
        -0.078786634,
        -0.032965753,
        0.022360912,
        0.012370768,
        0.069517635,
        -0.06794256,
        0.035108298,
        -0.019335406,
        0.047427155,
        -0.0027500102,
        0.021338332,
        -0.02817404,
        -0.017908305,
        -0.012590058,
        0.047914706,
        -0.041753206,
        0.017083809,
        -0.03069312,
        -0.006452188,
        0.052473247,
        0.016487358,
        0.016955068,
        -0.03678748,
        0.08069619,
        -0.032246836,
        -0.016139286,
        -0.032943826,
        0.025269104,
        0.02317144,
        0.020616695,
        -0.09917003,
        0.0749452,
        -0.103658155,
        0.0027698807,
        -0.016476551,
        -0.008453562,
        -0.0031145571,
        -0.027361348,
        0.033713263,
        -0.04798296,
        0.0018895635,
        0.009184487,
        0.08291782,
        -0.039845187,
        0.05542281,
        0.0005441681,
        0.03859006,
        -0.057004243,
        -0.049139246,
        0.026941286,
        -0.021696992,
        -0.046722915,
        0.030399092,
        -0.06307484,
        0.024602551,
        0.05410845,
        -0.0664596,
        0.01577382,
        -0.012335345,
        -0.031646516,
        0.029536093,
        -0.08396972,
        0.07214312,
        0.059898812,
        0.02435336,
        -0.0043443446,
        -0.06355041,
        0.010564461,
        -0.0035169513,
        -0.02079007,
        -0.08061384,
        -0.09630094,
        -0.009812978,
        0.0009157695,
        0.01906856,
        -0.0007333449,
        -0.57916856,
        0.012424566,
        -0.1121001,
        0.0664113,
        -0.16428903,
        0.03487487,
        0.015806999,
        0.024338568,
        0.08521961,
        -0.0212324,
        0.057667892,
        -0.12373118,
        0.034681197,
        0.016501646,
        0.035132013,
        -0.046937056,
        -0.021506958
      ]
    },
    {
      "id": "worldsfair-speaker-69",
      "conference": "worldsfair",
      "name": "Sam Lee",
      "role": "",
      "company": "NeoLabs",
      "twitter": "",
      "talks": [
        {
          "title": "FrontierCode needs Frontier Data Quality",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0042381887,
        0.05206844,
        -0.009255353,
        -0.010695414,
        -0.051247813,
        0.035370704,
        -0.04725907,
        0.035018086,
        0.100598894,
        -0.2580869,
        -0.033683114,
        -0.03200061,
        -0.040191147,
        0.094934165,
        -0.061716765,
        0.028710764,
        0.08714235,
        -0.065518565,
        -0.06300367,
        0.01108849,
        -0.08472006,
        -0.103630535,
        -0.0049284864,
        -0.014055129,
        -0.047456417,
        0.09245992,
        0.019277288,
        0.027453186,
        -0.09456521,
        0.49354306,
        -0.024483353,
        0.071012504,
        0.0082984185,
        -0.02673573,
        0.037181158,
        -0.045282315,
        -0.03691726,
        0.044644587,
        -0.013616649,
        0.031950574,
        -0.036442716,
        -0.012954613,
        -0.026719037,
        0.05164608,
        0.022053113,
        0.028157827,
        0.013801128,
        -0.067993745,
        -0.013317161,
        0.00077078334,
        -0.077919304,
        -0.024502328,
        -0.02181806,
        0.018613927,
        0.0025503822,
        0.0041894317,
        -0.012308464,
        -0.02611057,
        0.1071272,
        0.021948822,
        -0.16386163,
        -0.009918527,
        -0.03234889,
        0.0010329452,
        -0.025750885,
        -0.084368095,
        0.045153745,
        -0.053425886,
        -0.0012367079,
        -0.002808908,
        0.013161877,
        -0.004746825,
        -0.016210988,
        0.023243316,
        -0.1297338,
        -0.0060557313,
        0.014886749,
        0.085239835,
        -0.051401433,
        0.049812596,
        0.064937696,
        0.01358069,
        -0.043224838,
        -0.047004443,
        0.0026083568,
        -0.009939625,
        -0.07023282,
        0.0375019,
        -0.014868424,
        0.03542365,
        -0.0357968,
        -0.071472354,
        0.018177884,
        -0.0511943,
        -0.11541293,
        0.060574252,
        -0.040409837,
        0.017551564,
        -0.019739423,
        0.03960644,
        -0.063956894,
        -0.017577587,
        -0.02891111,
        -0.06661639,
        -0.07827486,
        -0.09938548,
        -0.040147047,
        -0.09029565,
        0.055294417,
        0.030651715,
        -0.027960364,
        -0.5837292,
        0.028957123,
        -0.09449476,
        0.037372705,
        -0.12517326,
        0.040680014,
        -0.035431035,
        0.05176214,
        0.093243934,
        0.01612806,
        0.06419616,
        0.014706044,
        0.04737577,
        -0.0011626632,
        0.048809536,
        0.04326277,
        0.0131721515
      ]
    },
    {
      "id": "worldsfair-speaker-70",
      "conference": "worldsfair",
      "name": "Uday Kiran Medisetty",
      "role": "",
      "company": "Uber",
      "twitter": "https://x.com/udaykiran",
      "talks": [
        {
          "title": "Agentic SDLC at Uber",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.06542929,
        0.03437169,
        0.025060248,
        0.11707051,
        0.039430182,
        0.0013477277,
        -0.054685503,
        0.052626062,
        0.058229256,
        -0.19862336,
        0.032725252,
        -0.053708646,
        0.04270376,
        -0.061925747,
        -0.045197025,
        -0.0013142636,
        0.09068327,
        -0.05381569,
        -0.029300544,
        -0.102528244,
        -0.00093906565,
        -0.1079904,
        0.0014624692,
        -0.07211185,
        -0.023429444,
        0.061632607,
        0.0049313707,
        0.05753121,
        -0.14260171,
        0.45582494,
        -0.05363187,
        -0.0051841782,
        0.054547098,
        -0.0418194,
        0.07956282,
        -0.08091818,
        -0.03678683,
        0.003008712,
        -0.0033554612,
        0.039919145,
        -0.03813319,
        0.033367835,
        -0.08134758,
        0.038371842,
        -0.016404234,
        0.020501263,
        -0.022108473,
        -0.0053190105,
        -0.024237536,
        -0.04454825,
        -0.09732052,
        0.002491431,
        0.016148206,
        -0.0467949,
        -0.023487598,
        0.025274327,
        -0.036711525,
        -0.058539484,
        0.07088951,
        -0.022624686,
        -0.059949283,
        0.005144856,
        -0.02060532,
        -0.04889909,
        -0.032476228,
        -0.08208211,
        0.10091626,
        -0.09471226,
        -0.0073892856,
        0.017288808,
        0.0009303434,
        0.044453878,
        -0.019121377,
        0.024789723,
        -0.055384308,
        -0.05320662,
        -0.015607096,
        0.11389191,
        -0.022553645,
        0.021298094,
        0.033827644,
        -0.004717977,
        -0.04655798,
        0.0012798848,
        0.020124918,
        -0.047501452,
        -0.06428693,
        -0.013284344,
        -0.069613196,
        0.078651704,
        0.039849903,
        -0.021938,
        -0.019660963,
        0.0066827154,
        -0.022824498,
        0.038226128,
        -0.06865871,
        0.05370526,
        -0.08287371,
        0.004580443,
        -0.074421324,
        -0.04287288,
        0.043110717,
        -0.0370015,
        -0.028247466,
        -0.025889428,
        -0.008794504,
        -0.055472992,
        -0.0053628385,
        -0.027070113,
        -0.014371816,
        -0.6360235,
        0.003835032,
        -0.017378457,
        -0.0031665573,
        -0.18360004,
        0.0045131873,
        -0.038926993,
        0.08790679,
        0.103973456,
        0.021897878,
        0.07501828,
        -0.009616306,
        0.05356075,
        -0.009579529,
        0.020824524,
        0.01342512,
        -0.014749248
      ]
    },
    {
      "id": "worldsfair-speaker-71",
      "conference": "worldsfair",
      "name": "Itamar Friedman",
      "role": "Co-Founder & CEO",
      "company": "Qodo",
      "twitter": "",
      "talks": [
        {
          "title": "The Last Human Code Review: Building Trust in AI-Generated Code",
          "description": "By the end of 2026, asking a human to review every pull request will be as optional as asking one to run every unit test manually. The tooling will be ready. The question is whether organizations are.\n\nIn this talk, Itamar Friedman, CEO of Qodo, explains why we are approaching the end of line-by-line human code review as a default requirement and explores what has to be true for teams to get there.\n\nThe barrier was never agentic AI capability. It was trust. And trust in automated review does not come from smarter models or faster feedback loops. It comes from systems that provide a trustworthy, concise and personalized proof-of-validation report. These systems are built on how engineering teams at specific organizations write their code: their own rules and standards, their PR history, their architecture decisions, their tribal knowledge that lives in comments and conversations and gets lost when engineers leave.\n\nItamar will walk through the shift from PR-by-PR review toward continuous, context-based code review and governance, and share a practical approach to making human code review optional.\n\nIf your team is shipping AI-generated code faster than humans can read it, join us for the discussion.",
          "day": "Day 2 — Session Day 1",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03751362,
        0.051311478,
        -0.042395875,
        -0.005459836,
        0.0065145777,
        0.012239052,
        -0.09746298,
        0.041846614,
        0.049369376,
        -0.2314259,
        -0.033957314,
        -0.03771552,
        -0.022481577,
        -0.03385281,
        -0.091018885,
        -0.010309243,
        0.1199844,
        -0.0084748855,
        -0.026372395,
        -0.10494648,
        -0.089112,
        -0.033634894,
        0.03504412,
        0.0783325,
        -0.029236712,
        0.10813025,
        0.009559237,
        0.05094218,
        -0.12118344,
        0.47633448,
        -0.1144855,
        -0.001400923,
        0.079508096,
        -0.047673237,
        0.037118964,
        -0.03985234,
        -0.07712977,
        -0.004052195,
        0.000051655057,
        0.053594526,
        -0.048702274,
        0.016800424,
        -0.06740842,
        0.07684567,
        -0.014442084,
        -0.08340759,
        -0.02046634,
        0.03287973,
        0.07413481,
        -0.050964598,
        -0.05798504,
        -0.04459128,
        0.043779936,
        0.024229221,
        0.052221518,
        0.023996968,
        -0.0039017035,
        -0.076676175,
        0.09044037,
        -0.03061801,
        -0.06464507,
        0.0068830727,
        0.026926722,
        0.006316209,
        -0.00044944047,
        0.006638661,
        0.027158566,
        -0.07573637,
        -0.022734731,
        -0.012351522,
        0.012653376,
        0.114469446,
        -0.02669644,
        -0.012904438,
        -0.04333491,
        -0.097311124,
        -0.026857791,
        0.08541374,
        -0.044271793,
        -0.05298073,
        0.03785685,
        0.009341865,
        -0.12550408,
        0.0033553797,
        0.028427921,
        -0.00082131394,
        -0.080212906,
        0.044101372,
        0.014553622,
        0.02830483,
        -0.017349062,
        -0.03977629,
        -0.029401891,
        0.07222465,
        -0.10948593,
        0.016752325,
        -0.076404564,
        0.03098756,
        -0.014812924,
        -0.027066348,
        -0.03715108,
        -0.031312037,
        0.023544434,
        0.0317978,
        0.016024064,
        -0.016927859,
        -0.008613152,
        -0.020227762,
        0.04704903,
        0.017042954,
        0.0021301724,
        -0.5815719,
        0.07854852,
        -0.0039332076,
        -0.05969916,
        -0.14408925,
        0.055192556,
        0.021468883,
        0.025778055,
        0.119262666,
        -0.0173761,
        0.10628397,
        -0.04073342,
        0.040939763,
        -0.019347034,
        0.08837799,
        0.022665085,
        0.031134084
      ]
    },
    {
      "id": "worldsfair-speaker-72",
      "conference": "worldsfair",
      "name": "Thomas Dohmke",
      "role": "CEO and co-founder",
      "company": "Entire",
      "twitter": "",
      "talks": [
        {
          "title": "What is the future of the SDLC?",
          "description": "# Fireside Chat Proposal: What is the future of the SDLC? The tools and processes we use to ship software — tickets, repos, pull requests, deployments — were designed for humans writing every line of code. That world is over. ## Discussion Points ### What does the \"next GitHub\" need to look like? Is there even such a thing? What would a platform built from the ground up for the AI era actually require? ### What's broken about the developer lifecycle as we know it The manual system of software production — from issues to git repositories to pull requests to deployment — was never designed for the era of AI. Where are the biggest seams showing? ### The review bottleneck: is code review a dying paradigm? Despite the boom in agent-generated code, the human developer still sits at the center of the pull request. Traditional code review assumes a human wrote every line and can explain every decision — but when an AI agent generates hundreds of lines from a brief prompt, reviewers face a fundamentally different challenge. How do teams adapt? ### The economics of agentic development In 2026, headcount can no longer be measured in salaries and benefits alone. Tokens are a real cost, with engineers reporting thousands of dollars a month in usage. How should engineering leaders think about this new variable? ### What \"agent-native\" tooling actually looks like The unit of work is shifting — from code as output to code-plus-context (prompts, reasoning, decisions) as the artifact. What does tooling built around that reality look like in practice?",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.009901298,
        0.033153642,
        0.041035302,
        0.04092109,
        0.06419691,
        0.035048585,
        -0.10622802,
        0.0064825364,
        0.10519305,
        -0.19211912,
        -0.055026095,
        0.00087225775,
        -0.0264536,
        -0.04596705,
        -0.10169373,
        -0.024372261,
        0.008668616,
        -0.046870124,
        0.002947674,
        -0.036079437,
        -0.022198832,
        -0.011022715,
        -0.008184748,
        0.10992614,
        -0.04577457,
        0.14828523,
        0.043242786,
        0.068499416,
        -0.14473502,
        0.4486211,
        -0.09487749,
        0.052944835,
        -0.0129562,
        -0.014464738,
        0.016346171,
        -0.033060268,
        -0.054103438,
        -0.014872103,
        0.05754917,
        0.10572984,
        -0.048991226,
        0.088502005,
        -0.06776461,
        0.011521015,
        -0.055359803,
        -0.049025733,
        -0.060675103,
        -0.04565502,
        0.0022099607,
        0.006297804,
        -0.0021746592,
        -0.024622703,
        0.041244052,
        0.0112213055,
        0.060613427,
        0.035025246,
        -0.018086856,
        -0.015110171,
        0.10659759,
        -0.007923759,
        -0.13286208,
        -0.015963208,
        0.058509074,
        0.054609027,
        -0.09836257,
        -0.075223126,
        0.017869165,
        -0.02357162,
        -0.028598517,
        -0.09299923,
        0.015670562,
        0.07779755,
        -0.12976651,
        0.059640326,
        -0.07764468,
        -0.08525823,
        0.030593898,
        0.07513313,
        0.015831629,
        -0.04995162,
        0.0022912677,
        0.0092142895,
        -0.08476589,
        -0.044603553,
        0.06506844,
        -0.030123014,
        -0.07051223,
        -0.020360056,
        0.028472384,
        0.06413786,
        0.01818527,
        0.001510855,
        -0.10428638,
        0.053421322,
        -0.053066615,
        -0.014154801,
        -0.05969413,
        0.0091403285,
        -0.004431349,
        0.005505113,
        -0.008422846,
        -0.014926932,
        -0.016733024,
        0.017216966,
        -0.026720587,
        0.014306816,
        0.05382929,
        -0.04839085,
        0.031891406,
        0.0020197,
        -0.041177273,
        -0.59133124,
        0.048147347,
        0.026992256,
        -0.043766133,
        -0.15203115,
        -0.025357608,
        0.00916966,
        0.08667674,
        0.084543936,
        -0.069994934,
        0.077020794,
        -0.042255577,
        0.038248833,
        0.013661833,
        0.07790156,
        0.0077307555,
        -0.00077264995
      ]
    },
    {
      "id": "worldsfair-speaker-73",
      "conference": "worldsfair",
      "name": "Shu Fang",
      "role": "",
      "company": "Two Sigma",
      "twitter": "",
      "talks": [
        {
          "title": "Tethered: Our Agents Are Us",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.12587109,
        -0.024728607,
        -0.015970787,
        0.011430025,
        0.026192432,
        -0.02854974,
        -0.05031639,
        -0.00056612224,
        0.03255494,
        -0.28130814,
        0.0022561487,
        0.023285985,
        0.013224371,
        0.04835007,
        -0.06684897,
        0.021073436,
        0.075981304,
        -0.06029246,
        -0.064906344,
        -0.027133705,
        -0.0026491368,
        -0.11521911,
        0.021033725,
        0.039540134,
        -0.015921568,
        0.08369275,
        0.00857262,
        0.008401723,
        -0.051084064,
        0.44535643,
        -0.06421074,
        0.072062925,
        0.076639265,
        -0.020158552,
        0.08267628,
        -0.07793516,
        -0.039484195,
        -0.043061685,
        0.04303235,
        0.06951944,
        -0.010362546,
        -0.03142883,
        -0.033993553,
        0.030671054,
        -0.07593172,
        -0.06395481,
        -0.034831293,
        -0.047606964,
        -0.085436806,
        0.08127534,
        -0.06711325,
        0.000068966256,
        0.021618627,
        -0.06663443,
        -0.03883212,
        -0.01575434,
        -0.019080834,
        -0.012999366,
        0.087074,
        0.02278053,
        -0.12941314,
        -0.0027485474,
        -0.005956554,
        0.033722036,
        0.023505138,
        -0.039267503,
        -0.026516378,
        -0.061901428,
        0.04657807,
        -0.006353648,
        -0.011875658,
        -0.038019653,
        -0.054981656,
        0.0655024,
        -0.13986138,
        -0.046172656,
        0.008209334,
        0.06620368,
        -0.027685363,
        -0.008059113,
        0.019826122,
        0.013882022,
        -0.122635975,
        -0.068524994,
        0.018459052,
        0.00015291545,
        -0.06916389,
        0.036676344,
        -0.0077100736,
        0.07095852,
        0.034247246,
        0.07442135,
        0.022442045,
        0.019775767,
        -0.1370428,
        0.0028065564,
        -0.11662714,
        0.089519195,
        -0.012986349,
        0.029366242,
        -0.013474097,
        -0.034121934,
        0.0033969318,
        0.019238934,
        -0.01315326,
        -0.07428064,
        -0.06746793,
        -0.053388488,
        0.15487653,
        -0.015243223,
        -0.036760397,
        -0.5821199,
        -0.030560799,
        0.01799295,
        -0.032628473,
        -0.105935425,
        -0.020938477,
        0.0067693433,
        0.06368444,
        0.08996269,
        -0.066791445,
        0.048943564,
        0.030187257,
        -0.008809055,
        -0.074794546,
        0.019424796,
        0.018901762,
        0.030773107
      ]
    },
    {
      "id": "worldsfair-speaker-74",
      "conference": "worldsfair",
      "name": "Benjamin Clavié",
      "role": "Member of Technical Staff",
      "company": "Mixedbread",
      "twitter": "https://x.com/bclavie",
      "talks": [
        {
          "title": "If we want them to do Knowledge Work, we need to design Knowledge Agents",
          "description": "It's tempting to assume that just like agents revolutionised coding, they will revolutionize other areas: legal, finance, advertising, and even medicine. All of those have in common that they are fundamentally knowledge work. And thankfully, humans have spent thousands of years searching for the best possible workflows for knowledge work. And yet, we seem to be disregarding all of these learnings, forcing every knowledge task into the shape that worked for coding. Today, we're going to talk about the history of knowledge work and how tools were co-designed to support it to understand how we should be building Knowledge Agents, themselves co-designed with their Knowledge Tools. This is key to avoiding falling into a \"good enough\" local optimum: think about legal clerking, a core part of the legal industry where information gathering and reasoning is performed to support the work of senior lawyers. The practice of clerking follows its own code, rules and best practices, which could not have feasibly emerged from studying software engineering: and similarly, there is no reason to believe knowledge agents could emerge from coding agents.",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09074328,
        -0.02489855,
        -0.039829813,
        0.02864948,
        0.02346688,
        -0.08894602,
        -0.043979272,
        0.0052469126,
        0.060528483,
        -0.21174118,
        -0.037147984,
        0.01660789,
        0.10222919,
        -0.0003201788,
        -0.068164356,
        -0.016773174,
        0.15410177,
        -0.058878634,
        -0.0681921,
        -0.07692408,
        -0.05725794,
        -0.036772057,
        -0.018546056,
        0.06920699,
        -0.07268865,
        0.09733824,
        0.02558462,
        0.031515077,
        -0.09756891,
        0.49522507,
        -0.1143509,
        0.02147054,
        0.05288301,
        0.00807125,
        0.03053813,
        -0.048430752,
        0.007968584,
        0.012343605,
        0.03296578,
        0.054377083,
        -0.040011507,
        0.036958322,
        -0.036261488,
        0.021973614,
        -0.02859346,
        0.030804742,
        -0.07406289,
        -0.019574763,
        -0.024927782,
        -0.013547901,
        -0.086026184,
        0.05728176,
        0.089680545,
        0.0050491313,
        0.04836513,
        -0.010978898,
        0.025405075,
        -0.04003744,
        0.06853145,
        -0.007206172,
        -0.010743474,
        -0.038658597,
        0.011058843,
        0.03204828,
        0.01769211,
        -0.028456138,
        0.06578538,
        -0.025546575,
        0.043778136,
        -0.064402275,
        0.06518225,
        0.049299628,
        -0.07232836,
        0.023070626,
        -0.086836345,
        -0.079065144,
        -0.02981994,
        0.044566035,
        -0.036039606,
        -0.0053910986,
        0.01376767,
        0.07751152,
        -0.046877984,
        0.0023224598,
        0.036379516,
        -0.03826217,
        -0.023516165,
        0.05657818,
        -0.044601083,
        0.023764381,
        0.08119796,
        0.012204012,
        0.0011379804,
        -0.0107145235,
        -0.074745536,
        -0.033855997,
        0.007583813,
        0.027391633,
        -0.015572905,
        0.018922027,
        0.0009652321,
        -0.039775576,
        0.08825765,
        -0.01926183,
        0.048691325,
        -0.049762435,
        -0.030138478,
        0.016839184,
        0.055799928,
        0.012411472,
        0.005601273,
        -0.61390346,
        -0.012186222,
        0.036141668,
        -0.0054282993,
        -0.1858507,
        0.056818925,
        0.04113967,
        0.01749743,
        0.093285754,
        -0.03576317,
        0.038460385,
        0.03408408,
        -0.0014481586,
        0.006770819,
        0.019375036,
        0.024337051,
        -0.002565941
      ]
    },
    {
      "id": "worldsfair-speaker-75",
      "conference": "worldsfair",
      "name": "Kim Maida",
      "role": "Head of Developer Relations & Founding GTM Engineer",
      "company": "Keycard Labs",
      "twitter": "https://x.com/DevRelUs",
      "talks": [
        {
          "title": "It's 10pm. Do You Know Where Your Agents Are?",
          "description": "Agents right now can sign legal contracts, run untethered, manage your dating profile, conduct financial transactions, and push code to production. Most agents have long-lived API keys and are dangerously overprivileged even when they're not making requests. In this talk, I'll demo how to solve the problem with the right access at the right time. You'll walk away knowing how to control agent access whether you're running coding agents from the CLI, building MCP servers, or connecting agents to third-party APIs.",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.022003971,
        0.07708019,
        0.085224934,
        0.018254468,
        0.025948534,
        -0.034892097,
        -0.05281113,
        -0.018745983,
        0.007922182,
        -0.27072918,
        0.0037952913,
        -0.036486764,
        -0.00080984796,
        0.0077482336,
        0.0041208807,
        -0.0013031262,
        0.053467434,
        -0.11694431,
        -0.009058256,
        -0.09597553,
        -0.06317394,
        -0.05246092,
        0.04329687,
        0.04817855,
        -0.05651556,
        0.054949686,
        0.004675859,
        0.106475174,
        -0.079922,
        0.45053166,
        -0.094490804,
        0.03728029,
        0.026050892,
        -0.012031803,
        0.03245571,
        -0.07157933,
        -0.057968237,
        -0.005861134,
        0.00095442426,
        0.053053346,
        -0.031720787,
        0.044746898,
        -0.065908715,
        0.03336482,
        -0.080760986,
        -0.022777375,
        -0.024034718,
        0.028009972,
        -0.00025886504,
        0.015101985,
        -0.038765892,
        -0.011933021,
        0.050720878,
        0.046956215,
        0.05688341,
        0.0054876744,
        0.011921335,
        0.010956314,
        0.09676449,
        -0.03174503,
        -0.08466891,
        0.014082288,
        -0.044824243,
        0.019555286,
        0.01526422,
        -0.11625335,
        0.025738612,
        -0.04297411,
        -0.002540326,
        -0.020367242,
        0.06605145,
        0.0835559,
        -0.03996309,
        0.023297133,
        -0.055309393,
        -0.14168295,
        -0.022829583,
        0.026502246,
        0.011993115,
        -0.03105763,
        0.004536375,
        0.06030655,
        -0.07556349,
        -0.09847911,
        0.036929887,
        -0.03486603,
        -0.111140095,
        0.025541412,
        -0.01085767,
        0.044694684,
        0.03526689,
        -0.03448082,
        -0.0893568,
        -0.0021040954,
        -0.14248708,
        -0.008443422,
        -0.04214012,
        0.028930912,
        -0.0043428466,
        0.023824863,
        -0.0010787526,
        -0.07156974,
        0.07611472,
        -0.017106282,
        0.014627366,
        0.02726163,
        0.021171022,
        -0.05292645,
        0.06905152,
        -0.031536646,
        -0.06395603,
        -0.60758924,
        -0.007517641,
        -0.033033706,
        0.010587865,
        -0.15080936,
        0.05751927,
        -0.030829025,
        0.06369754,
        0.06875387,
        -0.087942526,
        0.032304827,
        -0.039082043,
        -0.023698656,
        -0.031491514,
        -0.013203118,
        0.038688846,
        0.018842818
      ]
    },
    {
      "id": "worldsfair-speaker-76",
      "conference": "worldsfair",
      "name": "Neil Zeghidour",
      "role": "CEO",
      "company": "Gradium",
      "twitter": "https://x.com/neilzegh",
      "talks": [
        {
          "title": "Your Voice Agent is Just a Walkie-Talkie",
          "description": "Everyone says cascaded voice pipelines are dead and native speech models are the future. Yet production environments are still dominated by STT-LLM-TTS stacks. Reconciling the natural flow of native audio with the elite reasoning of a cascaded agent remains an unsolved systems problem. This talk dissects the brutal technical trade-offs behind that counterintuitive reality. We will break down why your voice agent is still stuck behaving like a walkie-talkie and map out the specific technical roadmap required to build full-duplex AI that actually works.",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        },
        {
          "title": "Everybody Gets a Digital Clone! (Part 1 of 3)",
          "description": "Walk out of this workshop with a deployed digital clone that makes your phone calls for you. We will skip the theory and immediately get our hands dirty wiring together OpenClaw, Twilio, and Gradium to build an autonomous voice agent on a live cellular network. You will tackle the hardest parts of real-time telephony: routing audio streams, handling human interruption, and killing latency. In 60 minutes, your AI will be ready to call restaurants for the daily special, book appointments, and actively negotiate on your behalf.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        },
        {
          "title": "Everybody Gets a Digital Clone! (Part 2 of 3)",
          "description": "Continuation of Neil Zeghidour's hands-on workshop on building a deployed digital clone for real-time phone calls using OpenClaw, Twilio, and Gradium.",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        },
        {
          "title": "Everybody Gets a Digital Clone! (Part 3 of 3)",
          "description": "Final continuation of Neil Zeghidour's hands-on workshop on building a deployed digital clone for real-time phone calls using OpenClaw, Twilio, and Gradium.",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        },
        {
          "title": "Voice is the universal interface",
          "description": "Language models give us the ability to create natural language, conversational, interfaces for computers. We are seeing a rapid shift among early adopters to using general language instead of traditional user interfaces for tasks like writing code and editing spreadsheets. Join the cofounders of Pipecat, Gradium, and Daily as we discuss the future of realtime voice and AI interfaces. Voice is the most efficient input mode for natural-language systems, and often the most efficient output mode, as well. But good voice interfaces require a very high degree of conversational facility, intelligence, task-specific reliability, and robustness to real-world realities like multiple speakers and background noise. There's a long history of voice interfaces in science fiction: Star Trek, Iron Man, Her. We'll use these depictions of computing possibilities as a jumping off point for talking about the ideal voice interface. How close are we to being able to build these interfaces with today's models, hardware, orchestration tooling, and UI libraries? What are the most promising research directions? What did the movies get wrong, now that we actually have experience building natural language, open-ended, voice systems?",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.026659109,
        0.05564745,
        0.05067184,
        0.010210053,
        -0.0061843656,
        0.010596019,
        -0.09627023,
        0.016937105,
        0.04625929,
        -0.25910872,
        0.03467919,
        -0.050525453,
        -0.013680565,
        -0.046865266,
        -0.118422516,
        0.040578585,
        0.06868607,
        -0.03521544,
        -0.04838482,
        -0.08176312,
        -0.02699127,
        -0.08018842,
        -0.053992648,
        0.007740636,
        -0.07426924,
        0.04878208,
        0.039744142,
        -0.04064697,
        -0.07533551,
        0.4165788,
        -0.060551774,
        -0.032458756,
        -0.006815961,
        -0.012587059,
        0.009135467,
        -0.06952748,
        -0.027804783,
        0.006224762,
        0.054659348,
        0.11912273,
        -0.039905824,
        0.044003323,
        -0.06679021,
        -0.038723342,
        0.031571094,
        0.0066552935,
        -0.053541936,
        0.055805035,
        0.0138800405,
        0.020470036,
        -0.03213819,
        0.04309746,
        0.03840154,
        0.024968632,
        0.0743142,
        0.027867597,
        0.006706198,
        -0.06407954,
        0.14922772,
        -0.021271572,
        -0.092319235,
        -0.046341743,
        0.031216241,
        0.06567583,
        0.003319441,
        -0.0028310237,
        -0.063357525,
        -0.036019,
        -0.022419447,
        -0.07784122,
        -0.04344838,
        0.011136624,
        -0.03857684,
        0.01621812,
        -0.042143438,
        -0.14575477,
        -0.017451582,
        0.04998838,
        -0.049631782,
        0.035979968,
        -0.05190785,
        0.06861559,
        -0.11188033,
        -0.05489259,
        0.05208717,
        -0.09895475,
        -0.06736444,
        0.10960328,
        -0.008336379,
        0.07235253,
        0.05294946,
        -0.019868074,
        -0.011016484,
        -0.00053777354,
        -0.010836496,
        -0.0056992774,
        -0.013389641,
        0.0043963366,
        0.07386396,
        0.03800293,
        0.042951055,
        -0.028570747,
        0.05223993,
        -0.02335903,
        -0.034271594,
        -0.053261213,
        -0.053672247,
        -0.005249684,
        0.017336095,
        0.007258483,
        -0.014181879,
        -0.64209867,
        0.0077504036,
        0.051997293,
        0.0027010185,
        -0.14355588,
        0.025738098,
        0.026867487,
        0.025760949,
        0.04658169,
        0.0023752467,
        0.025929427,
        0.018864209,
        0.042114094,
        0.0029607809,
        0.06866492,
        0.021005036,
        -0.061868064
      ]
    },
    {
      "id": "worldsfair-speaker-77",
      "conference": "worldsfair",
      "name": "Rayan Garg",
      "role": "Co-founder and CEO",
      "company": "Theta Software",
      "twitter": "https://x.com/RayanGarg",
      "talks": [
        {
          "title": "Theta Software",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.018743636,
        0.11329748,
        -0.004411436,
        0.027147483,
        0.077559076,
        -0.018550979,
        -0.09340391,
        0.014345705,
        0.02043771,
        -0.25018984,
        -0.058743227,
        -0.01166626,
        -0.0008845637,
        -0.0067289327,
        -0.07839568,
        0.02317363,
        0.022515643,
        -0.09814535,
        -0.03538573,
        -0.09511408,
        -0.049647328,
        -0.1119613,
        -0.03460407,
        -0.00064840383,
        -0.023041423,
        0.030581012,
        0.016333653,
        0.036060676,
        -0.07072844,
        0.43237332,
        -0.021742905,
        0.001643224,
        0.023349801,
        0.02570483,
        0.00029131927,
        -0.041486867,
        0.016663773,
        0.0053422786,
        0.03344591,
        0.07696301,
        -0.024646483,
        0.07407627,
        -0.058177456,
        0.053356573,
        0.02209373,
        0.089791894,
        -0.110200696,
        -0.0673676,
        0.0053828997,
        -0.0027946027,
        -0.0620448,
        -0.009920655,
        0.048170164,
        -0.03330379,
        -0.007563036,
        0.006675856,
        -0.09325068,
        -0.023677448,
        0.077374026,
        -0.025849365,
        -0.12694032,
        -0.033438332,
        0.019253984,
        -0.04248915,
        -0.013431296,
        -0.047387097,
        0.025543764,
        -0.02354352,
        -0.020734696,
        -0.008690135,
        0.049448006,
        -0.037619498,
        -0.043306224,
        0.09304782,
        -0.14104958,
        -0.032280385,
        -0.027936842,
        0.066662736,
        -0.007743439,
        0.022053694,
        0.055160828,
        0.01108697,
        -0.094194464,
        -0.026303776,
        -0.0055002626,
        -0.013684518,
        -0.00850907,
        0.07869863,
        0.041304257,
        0.12320266,
        -0.019853432,
        -0.07089238,
        -0.056588348,
        0.004145309,
        -0.034079015,
        0.027753152,
        -0.13086352,
        0.027938103,
        -0.045009047,
        0.008106911,
        0.043934662,
        -0.016354334,
        0.042834412,
        -0.06648081,
        -0.08869722,
        -0.02721097,
        0.053098917,
        -0.007489808,
        0.01445665,
        -0.024682818,
        -0.05681456,
        -0.5957601,
        0.028205466,
        0.028710017,
        -0.08971086,
        -0.16758966,
        0.046821386,
        -0.034896128,
        0.039125793,
        0.10946516,
        -0.06632558,
        0.11073757,
        -0.020239426,
        0.05597866,
        -0.048198204,
        0.057481397,
        0.027741494,
        0.050753534
      ]
    },
    {
      "id": "worldsfair-speaker-78",
      "conference": "worldsfair",
      "name": "Neha Singhal",
      "role": "",
      "company": "Uber",
      "twitter": "",
      "talks": [
        {
          "title": "Scaling Code Quality: Building uReview, Uber’s Multi-Agent Code Review Engine",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.036804788,
        0.014257067,
        0.0003221473,
        0.055818487,
        -0.04303544,
        0.012919044,
        -0.13905041,
        0.015906423,
        0.022434229,
        -0.22091821,
        0.030309262,
        -0.044263393,
        0.00349697,
        -0.046181,
        -0.07467566,
        -0.0039221835,
        0.059945837,
        -0.034508258,
        -0.0031241958,
        -0.07026052,
        -0.020013351,
        -0.05188289,
        0.065118834,
        -0.00889928,
        -0.07052487,
        0.02850228,
        0.014019797,
        0.014703165,
        -0.13675869,
        0.50736755,
        -0.09089409,
        -0.017796777,
        0.036965672,
        -0.04654356,
        0.051197845,
        -0.04308669,
        -0.0019023537,
        0.018550698,
        0.01910212,
        0.009050945,
        -0.01911323,
        0.023428263,
        -0.05179936,
        0.062244467,
        -0.042904504,
        -0.026118115,
        -0.027266858,
        -0.023952333,
        -0.033229914,
        -0.06390916,
        -0.04447589,
        -0.044112068,
        0.000803669,
        -0.092413835,
        0.033350557,
        0.034731023,
        -0.062069483,
        -0.07296311,
        0.09326301,
        -0.014941644,
        -0.109953985,
        0.00018601617,
        0.009534376,
        -0.023889957,
        0.020718737,
        -0.08236744,
        0.10628505,
        -0.11214078,
        0.009051641,
        0.04509431,
        0.04177382,
        -0.036362458,
        -0.05446648,
        0.07268621,
        -0.05630011,
        -0.08742652,
        0.0046988763,
        0.10260414,
        -0.021770496,
        0.062378913,
        0.010067896,
        0.009094265,
        -0.11690747,
        0.0056004594,
        0.05009095,
        -0.009132235,
        -0.050366715,
        0.01969896,
        -0.08792809,
        0.09753601,
        0.024283431,
        -0.02280932,
        0.008491116,
        -0.00248512,
        -0.023345415,
        0.022796286,
        -0.06914718,
        0.07545552,
        0.011815738,
        0.06898795,
        0.018062677,
        -0.071078256,
        0.016314656,
        0.019171773,
        -0.06509061,
        0.02274932,
        -0.056413066,
        -0.029559512,
        0.027332043,
        -0.00058898324,
        -0.023234572,
        -0.56996924,
        0.052319746,
        0.022086296,
        -0.03517095,
        -0.1376667,
        0.08288142,
        -0.056702614,
        0.027698115,
        0.121441536,
        0.0015396649,
        0.102191865,
        -0.06170118,
        -0.002605412,
        -0.0422087,
        0.024592448,
        0.06044163,
        0.00287524
      ]
    },
    {
      "id": "worldsfair-speaker-79",
      "conference": "worldsfair",
      "name": "Hursh Agrawal",
      "role": "Co-founder and CTO",
      "company": "The Browser Company",
      "twitter": "",
      "talks": [
        {
          "title": "Prototyping as Leadership: How a CTO Ships with AI Agents",
          "description": "I am a CTO and co-founder with a toddler, 15+ recurring meetings a week, 7 direct reports, and right now—7 open pull requests across two repos. Most engineering leaders eventually hit a wall where this kind of calendar tetris forces them to stop shipping code and start communicating solely through roadmaps. But what if AI agents didn't just act as coding assistants, but fundamentally restructured how executives use fragmented time to prototype the future? In this talk, I will share the exact multi-model workflows I use to plan with one model, implement with another, and build asynchronous play-and-feedback loops that fit perfectly between meetings. You will learn how to navigate code reviews for agent-assisted executive PRs, and leverage AI to shift your leadership style from telling your team what to build to showing them functional prototypes.",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0013696798,
        0.0909264,
        0.028190197,
        -0.0011930199,
        -0.033383735,
        -0.038414657,
        -0.13126867,
        -0.019174617,
        0.012352026,
        -0.1873474,
        -0.04691236,
        -0.025041463,
        0.02234003,
        -0.012109724,
        -0.077935725,
        -0.0015854469,
        0.057291392,
        -0.039465547,
        -0.0439576,
        -0.07686001,
        -0.05552004,
        -0.062251296,
        -0.061258055,
        0.07114569,
        -0.03390406,
        0.0996375,
        0.07392065,
        0.078064434,
        -0.12939042,
        0.46824747,
        -0.08693669,
        -0.0024182873,
        0.04084778,
        -0.029463068,
        0.06929444,
        -0.043959994,
        -0.050105292,
        -0.017266924,
        0.02810854,
        0.07396211,
        -0.018534666,
        0.06776006,
        -0.029266195,
        0.012234899,
        -0.050462935,
        -0.044667404,
        -0.048282947,
        -0.019364933,
        -0.03122917,
        0.027787194,
        -0.044714898,
        0.01751775,
        0.047074653,
        -0.008802072,
        0.032798827,
        0.048496272,
        0.065052085,
        -0.08759047,
        0.034710042,
        -0.016381722,
        -0.1318384,
        0.011032938,
        0.027268486,
        -0.011614115,
        0.017327515,
        -0.08907203,
        0.06046497,
        -0.0034313463,
        0.022519771,
        -0.018240504,
        0.00077896536,
        0.012991838,
        -0.05337556,
        0.036625218,
        -0.064542666,
        -0.10488944,
        0.0038624078,
        0.080324315,
        0.010553306,
        -0.019751221,
        0.030311307,
        0.01228959,
        -0.068875164,
        -0.06792715,
        0.08497559,
        -0.038972095,
        -0.08119159,
        0.031291563,
        0.02185305,
        0.08222634,
        -0.008236723,
        -0.02821153,
        -0.11108377,
        0.051863495,
        -0.05146271,
        0.06127783,
        -0.05526911,
        0.058412254,
        0.024252763,
        0.023683177,
        0.028234564,
        -0.043389015,
        0.0023159883,
        -0.027793057,
        0.0068260506,
        -0.036282245,
        0.059050053,
        0.015478234,
        0.059273686,
        -0.023158863,
        0.0025935324,
        -0.6224763,
        0.003630284,
        0.058620006,
        -0.032538448,
        -0.13839094,
        0.036070555,
        -0.011043928,
        0.067697324,
        0.077291235,
        -0.04811751,
        0.07657397,
        -0.019254152,
        0.010425018,
        -0.044908486,
        0.06273008,
        -0.010264754,
        0.06833623
      ]
    },
    {
      "id": "worldsfair-speaker-80",
      "conference": "worldsfair",
      "name": "Dat Ngo",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Your Agent Is Lying to You About Whether It Worked",
          "description": "Every span is green, every tool call returned cleanly, and the agent still regenerated the same plan 27 times before giving up invisible to any outcome metric, obvious in the trajectory. We pull up a real trace where the outcome looks healthy and the path is a disaster, then show Signal, our agent, surfacing it automatically: sweeping the project, ranking it above the noise, and linking straight to the offending trace with debugging evidence attached. The live version of the trajectory-over-outcomes argument, with a one-click path from \"something's wrong\" to \"here's exactly where.\"",
          "day": "Day 2 — Session Day 1",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.049168684,
        0.052686803,
        0.01813308,
        0.05044683,
        -0.0066262544,
        -0.0035184207,
        -0.060466256,
        0.008273007,
        0.027410312,
        -0.29696342,
        -0.0007850291,
        0.042810053,
        0.00567662,
        -0.008981908,
        -0.051775225,
        -0.028500706,
        0.091253385,
        -0.06283106,
        -0.08323981,
        -0.064910166,
        -0.06866618,
        -0.042245995,
        -0.023371287,
        0.068454914,
        0.022606757,
        0.084927104,
        0.0211458,
        0.10513374,
        -0.07225544,
        0.4293418,
        -0.06563264,
        0.018912476,
        0.0737146,
        -0.064070374,
        0.032979846,
        -0.09189117,
        -0.016034346,
        -0.015512384,
        0.009660438,
        0.07675969,
        0.015907155,
        0.046235945,
        -0.017390857,
        0.03637684,
        -0.08067107,
        -0.024431882,
        -0.11042617,
        -0.00017134394,
        -0.007359685,
        -0.023097597,
        -0.06832796,
        -0.008233256,
        -0.0048495973,
        0.01816026,
        0.027205097,
        0.07474192,
        0.019099828,
        -0.040135954,
        0.07623593,
        -0.0055615255,
        -0.14260261,
        0.0687969,
        0.039130323,
        -0.022137972,
        0.023409136,
        -0.09484525,
        0.036764145,
        -0.047675453,
        0.008710554,
        -0.06733662,
        0.01121101,
        -0.0018519901,
        -0.06418187,
        -0.01336098,
        -0.03608732,
        -0.15425096,
        -0.06609044,
        0.060715325,
        -0.021764189,
        -0.015207528,
        -0.052460346,
        0.07582121,
        -0.07164042,
        -0.04671025,
        0.08555365,
        0.002732221,
        -0.08439826,
        0.056838237,
        -0.037497636,
        0.06335864,
        0.032563966,
        -0.02617281,
        0.0001379849,
        0.043978605,
        -0.06225419,
        0.030709002,
        -0.040924445,
        0.06178213,
        -0.06756007,
        0.056626383,
        -0.054019373,
        -0.0071090385,
        0.038721304,
        -0.026663827,
        0.0016292128,
        -0.03166748,
        0.018462857,
        -0.008243843,
        0.07222765,
        0.08034141,
        -0.00034621602,
        -0.6050572,
        0.009032784,
        -0.017440218,
        -0.020572752,
        -0.16960251,
        0.028299006,
        0.041175082,
        0.044401687,
        0.036131255,
        -0.017391449,
        0.09122129,
        -0.0076977797,
        -0.029203868,
        -0.03427639,
        0.015629563,
        0.040281378,
        0.019448709
      ]
    },
    {
      "id": "worldsfair-speaker-81",
      "conference": "worldsfair",
      "name": "Zach Lloyd",
      "role": "Founder and CEO",
      "company": "Warp",
      "twitter": "https://x.com/zachlloydtweets",
      "talks": [
        {
          "title": "Self-Improving software factories: The new open source model\"",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        },
        {
          "title": "The Intelligence Infrastructure You Should Own",
          "description": "A pattern has emerged in 2026: many AI companies now offer some version of a managed agent platform. They saw the same problem: agents running on laptops do not scale, teams need central visibility, enterprise needs governance. They got a lot right. Cloud execution, event triggers, parallel agents, a centralized management layer. This category is real and it is shipping. But after talking to dozens of engineering leaders at leading enterprises, I think the current approaches share a structural gap. They focused on the orchestration layer. Almost none of them addressed what sits beneath it. Here is the distinction that matters. The orchestration service that coordinates your agents can run as a shared service, the way any SaaS control plane does. That is normal and fine. But the intelligence your agents accumulate over time is a different category entirely. That is the memory they build about your codebase, your team's institutional knowledge, and your historical decisions. That is what compounds in value over months and years. And in the current generation of managed platforms, that intelligence lives in the vendor's infrastructure, not yours. Stop using the platform, and it walks out the door. Switch harnesses, and you start from zero. The conversation around this is growing. Harrison Chase has been [writing about agent memory design](https://x.com/hwchase17/status/2042978500567609738). Andrej Karpathy's tweet on this topic went viral among developers, which signals that this gap is being felt across the engineering community. What has not yet shipped is a platform-level implementation where the intelligence layer belongs to you, not the vendor. What good intelligence infrastructure looks like has several layers, not just one or two. ![oz_diagram.png](attachment:a71d92a8-b456-42ac-b3bf-38d0e593aee2:oz_diagram.png) Above: orchestration. A control plane that governs which models and harnesses run. Coordination across providers, not locked inside one. Evaluation and observability that benchmarks harnesses against each other. Enterprise governance that works on your infrastructure, not the provider's cloud. Beneath: the foundation you own. Persistent memory that compounds across sessions, agents, and harnesses. Context architecture built from your codebase, your team's decisions, your workflows. Stored on your infrastructure, exported if you leave, accessible via API by any agent you run. Human-in-the-loop review built into the process so your institutional knowledge is curated, not just accumulated. The difference in outcome: agents that know your business, not agents that know a pre-defined process owned by whoever built the harness. This is also how you future-proof your investment. The memory layer you build today is the substrate that will make reinforcement learning approaches and future model architectures more effectively. When your intelligence infrastructure belongs to you, you can plug in tomorrow's capabilities without starting over. I will share a story that crystallized this for me. A VP of AI at a large US financial institution came to a meeting with me having independently built a home governance system for his internal agent sessions: agents must register with an orchestrator, submit a specification, and have it validated by another agent before execution begins. He was not doing this because he is a systems nerd. He was doing it because the available platforms did not give him what he needed. He was building his own foundation layer, by hand. That is the pattern. By the time of this conference, Warp will have several months of customer insights from our own platform. I will share what we learned, including what surprised us, and the practical framework for making build vs. buy decisions at each layer of the stack.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.012702938,
        0.083142705,
        0.058587372,
        0.01889262,
        0.034497112,
        -0.040519502,
        -0.067065336,
        -0.028509414,
        0.021357879,
        -0.25297493,
        -0.062183015,
        -0.04613891,
        0.056729764,
        -0.061836887,
        -0.027905062,
        0.002385445,
        0.06394705,
        -0.05442162,
        -0.083728254,
        -0.077621795,
        0.0019644138,
        -0.07201747,
        -0.011564103,
        0.011223346,
        -0.013404597,
        0.0600474,
        0.03297007,
        0.062507175,
        -0.1229284,
        0.43466663,
        -0.06761169,
        -0.0035514608,
        0.031063257,
        -0.0006296972,
        0.02802804,
        -0.024366165,
        -0.030406302,
        -0.019496499,
        -0.008608056,
        0.027047053,
        -0.017626604,
        0.048602,
        -0.060536783,
        -0.044918273,
        -0.042609666,
        -0.021836879,
        -0.06886149,
        -0.00847479,
        -0.040960252,
        -0.042596396,
        -0.066493884,
        -0.0035883116,
        0.11705963,
        0.039032232,
        0.057098698,
        0.00006356594,
        -0.010709728,
        -0.08386346,
        0.09868855,
        -0.07933187,
        -0.116750844,
        -0.013322908,
        0.045235533,
        0.11696452,
        -0.0018399516,
        -0.032707896,
        0.04817842,
        -0.049845647,
        -0.019506406,
        -0.08969351,
        0.026207794,
        0.036929388,
        -0.025682453,
        0.014123322,
        -0.12632862,
        -0.11446192,
        -0.04492368,
        0.10026217,
        0.024487343,
        -0.019375885,
        0.026808703,
        0.061557524,
        -0.07702722,
        -0.0017086797,
        0.011395383,
        0.01884506,
        -0.043695815,
        0.06348484,
        -0.06161794,
        0.079300165,
        0.04293638,
        -0.022925315,
        -0.044005234,
        0.030790407,
        -0.08916852,
        0.008435453,
        -0.06018808,
        0.06559496,
        0.009978281,
        0.030931216,
        -0.05798887,
        0.0037652717,
        0.060191363,
        -0.026052715,
        0.002105302,
        -0.061517045,
        0.028971916,
        0.0071763243,
        0.043727476,
        0.01637148,
        -0.052788127,
        -0.61287546,
        -0.046196435,
        0.044305895,
        0.025274841,
        -0.14817475,
        0.091260575,
        0.003760464,
        0.068029016,
        0.07069183,
        -0.0076686423,
        0.058956932,
        -0.049030688,
        0.063525446,
        0.034994822,
        0.053613167,
        0.022692662,
        -0.0026142627
      ]
    },
    {
      "id": "worldsfair-speaker-82",
      "conference": "worldsfair",
      "name": "Jean-Denis Greze",
      "role": "Co-Founder & CEO",
      "company": "Town",
      "twitter": "https://x.com/jgreze",
      "talks": [
        {
          "title": "Agents' next frontier: agent-to-agent and network effects",
          "description": "MCP v. CLI was about how agents talk to tools. That’s not settled (but we’re camp MCP… mostly). Almost nothing has settled how agents talk to each other - and that's where the next wave of value (and network effects and virality) lives. At Town we run a personal AI agent in production inside real people's inboxes, calendars, and Slack, and we've built agent-to-agent (A2A) on our platform: 1:1 A2A messaging, agents that carry a short bio of one another, HITL when sensitive data is shared or write actions are involved, and early tests around 1:N A2A. I’ll talk about the why, the opportunity, and the production architecture underneath. Audience takeaway: a concrete mental model for building multi-agent systems on top of the data and surfaces users already live in, plus our learnings on early failure modes to avoid.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.003953859,
        0.027514914,
        -0.015249068,
        0.0058573447,
        0.031006595,
        -0.03553361,
        -0.08288669,
        0.016664488,
        -0.0070950952,
        -0.259699,
        0.01500285,
        -0.09046652,
        -0.025039965,
        -0.03731986,
        -0.10410931,
        -0.0012549054,
        0.05552603,
        -0.07011877,
        -0.078731075,
        -0.07293674,
        -0.019296722,
        -0.0987305,
        -0.028982826,
        -0.052299492,
        0.0069782087,
        0.13518605,
        0.03312188,
        0.047146413,
        -0.09977467,
        0.43780658,
        -0.050156616,
        -0.010244973,
        0.024268962,
        -0.00971081,
        0.02744541,
        -0.08062325,
        0.027721655,
        -0.025714288,
        0.015002271,
        0.081373036,
        0.019936977,
        0.041842647,
        -0.06978151,
        0.014723289,
        -0.0677845,
        0.017431177,
        -0.056661535,
        0.039921034,
        -0.04365244,
        -0.0066481438,
        -0.07922071,
        0.046118896,
        0.0025080955,
        -0.0037139133,
        0.047606036,
        -0.00095967506,
        0.063319884,
        -0.0501028,
        0.09967179,
        -0.011654532,
        -0.11636411,
        -0.03166642,
        0.03181806,
        0.04806312,
        0.038799677,
        -0.05549354,
        -0.027755985,
        -0.051677007,
        0.032404963,
        -0.1281041,
        0.041807767,
        0.025007524,
        -0.0855438,
        -0.019613251,
        -0.08889404,
        -0.12654603,
        0.03182561,
        0.07156944,
        -0.006268765,
        -0.0131334765,
        0.005439699,
        0.046154045,
        -0.026890067,
        -0.061372835,
        0.038587034,
        -0.04008738,
        0.004631375,
        0.025770303,
        0.0037703575,
        0.011570177,
        0.06740859,
        0.0065267803,
        -0.04434111,
        0.03959472,
        -0.054525055,
        0.005640896,
        -0.029585134,
        0.08805685,
        0.038095307,
        0.0072679874,
        0.0038936168,
        -0.062191322,
        0.046695672,
        -0.013360624,
        -0.009012627,
        -0.00939823,
        0.040468343,
        -0.048818808,
        0.0391869,
        0.003080022,
        -0.008985054,
        -0.6286821,
        -0.0115188565,
        0.043420184,
        0.04392638,
        -0.15863374,
        0.020665143,
        0.07022469,
        0.035049792,
        0.061256357,
        -0.06247259,
        0.07615788,
        -0.011492786,
        0.041901715,
        -0.019700501,
        0.06949197,
        0.0069460827,
        -0.01420859
      ]
    },
    {
      "id": "worldsfair-speaker-83",
      "conference": "worldsfair",
      "name": "Simon Eskildsen",
      "role": "Co-Founder & CEO",
      "company": "turbopuffer",
      "twitter": "https://x.com/Sirupsen",
      "talks": [
        {
          "title": "How to Connect AI to Billions of Legal Documents",
          "description": "Legora’s foundational engineering challenge is connecting frontier LLMs to billions of legal documents so the models can efficiently solve end-to-end legal workflows without burning extra tokens. We’ll share the retrieval architecture we built with turbopuffer that achieves: 1. Strict data isolation across millions of legal cases in a very security-conscious domain 2. Predictable search performance (<100ms p90 latency) on large contexts 3. High retrieval quality (95%+ recall@10) with fewer agent loops We’ll retrospect on two architectures that failed to achieve all 3 (and why), and the key design factors that make the current solution work at our scale. Practical takeaways include: - How to evaluate per-tenant vs shared-index retrieval under strict data isolation - How to efficiently index and retrieve context to maximize relevance per input token - How to build a highly intelligent AI application when your inference budget is constrained",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06278374,
        0.017987391,
        -0.04587303,
        0.017414326,
        -0.011037251,
        -0.023697209,
        -0.09002427,
        -0.0044989865,
        0.09014315,
        -0.21231212,
        -0.0344186,
        -0.056523237,
        0.004284964,
        -0.0095657455,
        -0.087569356,
        0.021829532,
        0.11580087,
        -0.050041024,
        -0.06607111,
        -0.019384695,
        -0.021801347,
        -0.036815844,
        -0.028869603,
        0.07567978,
        -0.068341605,
        0.07062144,
        -0.03989911,
        0.0472458,
        -0.054716133,
        0.47376043,
        -0.14402162,
        0.023953473,
        0.0016080717,
        -0.034591965,
        -0.0191055,
        -0.06405808,
        -0.012419107,
        0.006013545,
        0.053194106,
        0.07741531,
        0.0012843142,
        0.07141477,
        -0.063575886,
        0.06748147,
        -0.013481784,
        -0.03420998,
        -0.054569148,
        0.07370905,
        0.0433212,
        0.035710923,
        -0.08545728,
        0.071495935,
        0.05548436,
        0.0043215025,
        0.03742978,
        0.064880654,
        0.04188568,
        -0.0621656,
        0.12359801,
        -0.078026496,
        -0.079616904,
        -0.09384231,
        0.011141236,
        -0.0050193844,
        -0.002079422,
        -0.022817234,
        0.033012923,
        0.016011296,
        0.051784955,
        -0.0781282,
        0.021546042,
        0.0024826012,
        -0.062359042,
        0.0028041538,
        -0.044358246,
        -0.11387564,
        -0.04262558,
        0.07001567,
        0.0002461993,
        -0.03478707,
        0.024064885,
        0.03833479,
        -0.031726215,
        -0.046969026,
        0.105989955,
        0.000099590936,
        -0.058537036,
        0.029528445,
        -0.0356302,
        0.1314978,
        0.007366312,
        0.021768663,
        -0.058742166,
        -0.0047809635,
        -0.07173723,
        0.037431937,
        -0.017820433,
        0.07163758,
        0.014847145,
        0.014040276,
        -0.030491289,
        0.04018329,
        0.06089285,
        0.017014448,
        -0.0356695,
        0.029176176,
        0.07388628,
        -0.017241098,
        0.10344335,
        -0.029347302,
        -0.04998428,
        -0.59957594,
        -0.033577327,
        0.03443354,
        0.011928375,
        -0.15081567,
        -0.002218515,
        -0.049044784,
        0.014841767,
        0.0459105,
        0.005571313,
        0.052421056,
        -0.0015429653,
        -0.036453847,
        -0.062307317,
        0.030488994,
        -0.02457757,
        0.040041447
      ]
    },
    {
      "id": "worldsfair-speaker-84",
      "conference": "worldsfair",
      "name": "Jacob Lauritzen",
      "role": "CTO",
      "company": "Legora",
      "twitter": "",
      "talks": [
        {
          "title": "How to Connect AI to Billions of Legal Documents",
          "description": "Legora’s foundational engineering challenge is connecting frontier LLMs to billions of legal documents so the models can efficiently solve end-to-end legal workflows without burning extra tokens. We’ll share the retrieval architecture we built with turbopuffer that achieves: 1. Strict data isolation across millions of legal cases in a very security-conscious domain 2. Predictable search performance (<100ms p90 latency) on large contexts 3. High retrieval quality (95%+ recall@10) with fewer agent loops We’ll retrospect on two architectures that failed to achieve all 3 (and why), and the key design factors that make the current solution work at our scale. Practical takeaways include: - How to evaluate per-tenant vs shared-index retrieval under strict data isolation - How to efficiently index and retrieve context to maximize relevance per input token - How to build a highly intelligent AI application when your inference budget is constrained",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.048058204,
        0.025562238,
        -0.0036322854,
        0.028256426,
        -0.0032524937,
        -0.072575614,
        -0.08926448,
        0.06567514,
        0.07222966,
        -0.2367017,
        -0.034907844,
        -0.049846467,
        0.01900693,
        0.02492467,
        -0.06254239,
        -0.022655185,
        0.18144065,
        -0.03915564,
        -0.08028097,
        -0.06961367,
        -0.015696911,
        -0.069613725,
        -0.052865107,
        0.036176465,
        -0.10213707,
        0.08630902,
        -0.04476629,
        0.018042875,
        -0.042856917,
        0.46469283,
        -0.10372521,
        0.050934926,
        0.006138386,
        -0.03580197,
        -0.008652138,
        -0.018537922,
        -0.0018451896,
        -0.022206603,
        0.043148644,
        0.07157866,
        -0.019036885,
        0.057259545,
        -0.043666217,
        0.051344246,
        -0.0059039947,
        -0.00095791527,
        -0.07123398,
        0.07668035,
        0.040571667,
        0.0017105854,
        -0.090565726,
        0.0010751506,
        0.03253392,
        0.002329009,
        0.025182351,
        0.044660766,
        0.08043376,
        -0.04221929,
        0.11812844,
        -0.010286299,
        -0.08300041,
        -0.118782304,
        -0.029408732,
        -0.03248989,
        0.015678566,
        -0.056186367,
        0.015428881,
        0.046168294,
        0.03280101,
        -0.075994074,
        -0.01015719,
        -0.014709301,
        -0.060987998,
        -0.027487716,
        -0.06013426,
        -0.083255336,
        -0.05580844,
        0.086557776,
        0.02239483,
        -0.0008119785,
        0.0065369294,
        0.06514203,
        -0.030897386,
        -0.04097057,
        0.10184731,
        -0.022936862,
        -0.06608021,
        0.044331297,
        -0.04872174,
        0.11986745,
        0.026508333,
        -0.047187474,
        -0.0351361,
        -0.033312716,
        -0.08383267,
        0.09113398,
        -0.0377976,
        0.051057275,
        0.03942468,
        0.048515934,
        -0.04215811,
        0.0072465087,
        0.045442775,
        0.040305045,
        -0.01918959,
        0.03016013,
        0.05487853,
        0.026035761,
        0.10026966,
        0.008201932,
        -0.06053348,
        -0.5886728,
        -0.026886342,
        0.020230724,
        0.018652571,
        -0.15284228,
        0.018894771,
        -0.051794376,
        -0.0037076655,
        0.028432718,
        0.0011528782,
        0.016910886,
        -0.008379451,
        -0.000863559,
        -0.074580684,
        0.017632972,
        0.016416268,
        0.018186834
      ]
    },
    {
      "id": "worldsfair-speaker-85",
      "conference": "worldsfair",
      "name": "Moritz Johner",
      "role": "Platform Architect / Senior Software Engineer",
      "company": "Form3",
      "twitter": "",
      "talks": [
        {
          "title": "We Gave an Agent Production Code Access and Then Tried to Sleep at Night",
          "description": "We let an agent touch production code to fix CVEs. That is either automation or a supply chain incident, depending on how honest your architecture is. PatchPilot started simple: find vulnerable dependencies, patch them, open a PR, let CI prove the fix, move on. Then reality showed up. The agent needed repository access, CI logs, credentials, and a Docker socket. Without that, it was useless. With it, every security reviewer in the room had a point. This is the production case study: what we gave the agent, what we refused, what infosec pushed back on, and where they were right. We will cover scoped permissions, constrained PRs, audit trails, approval gates, CI evidence, credential boundaries, and the gap between \"it generated a patch\" and \"we can defend this change.\" Agentic remediation is not just developer productivity. It is a new participant in your software supply chain.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.0733601,
        -0.027891211,
        0.048634365,
        -0.024020523,
        0.03599893,
        -0.029101953,
        -0.05396015,
        -0.038811933,
        0.08497113,
        -0.22346964,
        -0.09774251,
        0.021589352,
        0.039760727,
        -0.012000189,
        -0.07113972,
        -0.07745224,
        0.080118105,
        -0.05896981,
        -0.037168216,
        -0.057504397,
        -0.07045545,
        -0.031661958,
        0.015697476,
        -0.025144525,
        -0.0058931923,
        0.09114047,
        -0.028809834,
        0.01626164,
        -0.10100112,
        0.5165119,
        -0.074410066,
        0.0507914,
        0.033380184,
        -0.11096257,
        0.10472216,
        -0.04952043,
        -0.07990066,
        -0.0039274655,
        -0.0014251397,
        0.02747619,
        -0.0045667263,
        0.020864798,
        -0.095540464,
        0.046208896,
        -0.058391713,
        -0.024901671,
        -0.029216876,
        0.0710533,
        -0.021446431,
        -0.02354047,
        -0.10302536,
        0.04537392,
        0.043135375,
        -0.020369606,
        0.05048976,
        0.019350592,
        0.070503905,
        -0.040971458,
        0.06857813,
        -0.0444307,
        -0.06142269,
        -0.076785624,
        0.0053624986,
        0.019928122,
        -0.03199864,
        -0.046099514,
        0.07385688,
        -0.04285547,
        0.014960418,
        0.029575435,
        0.07875395,
        0.07879119,
        -0.05269588,
        0.007160674,
        -0.023552692,
        -0.09028092,
        -0.019902945,
        0.011964103,
        -0.014720545,
        0.018941147,
        -0.0104221385,
        0.056765206,
        -0.072378665,
        -0.043779273,
        0.0282414,
        0.033533942,
        -0.06373767,
        0.007771641,
        0.010899443,
        0.067617625,
        0.014819619,
        0.04052237,
        -0.033135843,
        0.13595994,
        -0.06406865,
        0.00858953,
        -0.033050507,
        -0.0068410086,
        0.006781939,
        0.025656778,
        -0.016319131,
        -0.03758531,
        -0.018986942,
        -0.023593143,
        0.033672865,
        -0.0027871972,
        -0.01341726,
        -0.0049347896,
        0.017824693,
        0.039273333,
        -0.04275619,
        -0.5835038,
        -0.03365708,
        -0.034237415,
        0.0009439785,
        -0.17878906,
        0.01894884,
        0.011482695,
        -0.0058642966,
        0.10816661,
        -0.038377408,
        0.04550789,
        -0.06845419,
        -0.010857912,
        -0.032180864,
        0.0041498793,
        0.029986223,
        0.01597228
      ]
    },
    {
      "id": "worldsfair-speaker-86",
      "conference": "worldsfair",
      "name": "Paula Rambles",
      "role": "",
      "company": "",
      "twitter": "https://x.com/paularambles",
      "talks": [
        {
          "title": "Tolan: Voice-First AI Companion",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.00041597162,
        0.070151515,
        -0.043053187,
        0.07444836,
        0.025351802,
        -0.03699009,
        -0.052879397,
        0.024308937,
        0.011823718,
        -0.2522687,
        0.0011910297,
        -0.036538202,
        -0.03199441,
        0.0014398764,
        -0.03446677,
        0.01682136,
        0.048618373,
        -0.075603195,
        -0.10709317,
        -0.097621486,
        -0.03771491,
        -0.07756065,
        -0.036854476,
        0.090487964,
        -0.09756058,
        0.030421559,
        -0.030896628,
        0.048297025,
        -0.070605405,
        0.42970115,
        -0.08326448,
        0.07426784,
        -0.002212882,
        -0.021670612,
        0.04383729,
        0.015048632,
        -0.0024280539,
        -0.05473418,
        0.012536014,
        0.07549646,
        -0.023347806,
        0.06474133,
        -0.032893304,
        0.009841106,
        0.031502865,
        -0.048685484,
        -0.068172134,
        0.0037586093,
        0.026870107,
        0.002035679,
        0.001865325,
        0.030225703,
        -0.023663806,
        0.008117596,
        0.052067492,
        0.013965976,
        -0.016428923,
        -0.042901915,
        0.051650453,
        -0.032845475,
        -0.09305159,
        -0.003542904,
        0.03201021,
        -0.0090248175,
        -0.012338617,
        -0.04993466,
        0.043640666,
        -0.08675326,
        0.00684572,
        -0.08784042,
        -0.0663975,
        0.03322975,
        -0.06553606,
        0.0635421,
        -0.06434161,
        -0.10140721,
        -0.051711753,
        0.06782492,
        0.04350425,
        -0.019494059,
        0.02642197,
        0.04803235,
        -0.08621643,
        -0.096883014,
        -0.03163462,
        -0.0030236673,
        -0.13492225,
        0.06210871,
        -0.018198086,
        0.042861544,
        -0.006313371,
        -0.022637174,
        -0.049531788,
        0.045687236,
        0.031731173,
        0.021046959,
        -0.025779987,
        0.055924833,
        -0.058009245,
        0.008722581,
        -0.006239172,
        -0.033453465,
        -0.051929418,
        -0.033706177,
        0.003183148,
        0.022549171,
        -0.01037641,
        -0.0054390263,
        0.059025083,
        0.020453328,
        -0.009806576,
        -0.6471543,
        0.02643143,
        -0.021287844,
        -0.023345824,
        -0.17052,
        -0.03160904,
        -0.018991409,
        0.08532655,
        0.08312723,
        0.009321051,
        0.048117172,
        -0.0021198494,
        -0.028687045,
        0.010912766,
        0.07150225,
        0.0811528,
        0.015818248
      ]
    },
    {
      "id": "worldsfair-speaker-87",
      "conference": "worldsfair",
      "name": "Varun Singh",
      "role": "Pre-Training Lead",
      "company": "Arcee AI",
      "twitter": "",
      "talks": [
        {
          "title": "The Base Model is Dead",
          "description": "It's a common belief that large language models are trained to be a good model of human web-text, and thus base models are \"mirrors\" of what we see on the internet. Historically, this was largely true, but no modern base model truly reflects the internet in the way that GPT-3 once did. Instruction data along with synthetic reasoning traces are moving earlier and earlier into the training pipeline, and \"mid-training\" has emerged as a new stage to accommodate longer datapoints that more concretely resemble downstream capabilities. As a result, pre-training no longer has the goal of creating a linguistic prior, but instead has the additional goals of baking in behavior and more atomic skills into the trained \"base\" model. Between this shift in what a base model is and the blurring of the lines between the different stages of model training, it's an open question as to what the best approach is here (at least outside the walls of the big labs). But I believe that the role we view the base model playing will continue to shift as we're pulled forward through new phases of model capabilities.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.043222226,
        0.10537373,
        0.020230152,
        0.016017387,
        -0.0044976687,
        0.014952657,
        -0.08029301,
        -0.025358379,
        0.036605775,
        -0.2233328,
        0.0052866046,
        -0.079442516,
        0.015402775,
        -0.05736651,
        -0.069963865,
        0.03961452,
        0.09753802,
        -0.07258044,
        -0.07580407,
        -0.066588506,
        0.03237869,
        -0.07616964,
        0.005115396,
        0.07371174,
        -0.033747375,
        0.064974025,
        -0.026010914,
        0.0009086262,
        -0.11672493,
        0.43124354,
        -0.09421155,
        -0.016788857,
        -0.040069763,
        0.014387576,
        0.048787355,
        -0.06580707,
        0.008579876,
        0.050377235,
        0.029285334,
        0.055903062,
        0.013087072,
        -0.009898426,
        -0.011355566,
        0.03929472,
        0.013899915,
        -0.052194975,
        -0.08114557,
        0.043469276,
        0.011438762,
        -0.05615053,
        -0.05785655,
        -0.017031897,
        0.04380382,
        0.03168645,
        0.042851858,
        0.0666121,
        0.04851265,
        -0.081873626,
        0.07488862,
        -0.06451874,
        -0.15091577,
        0.040998794,
        0.046097953,
        0.026451496,
        0.0019632587,
        -0.08117652,
        0.035963148,
        -0.077705055,
        -0.040442582,
        0.0037487827,
        0.0004803224,
        0.031129407,
        -0.091048166,
        0.016770655,
        -0.032329194,
        -0.112859204,
        -0.02509881,
        0.04907401,
        -0.0514109,
        0.03493265,
        -0.03893112,
        -0.025367578,
        -0.008966922,
        -0.015592626,
        0.099125415,
        -0.007560339,
        -0.042829007,
        0.105820686,
        -0.062836684,
        0.08091883,
        0.045861136,
        -0.005952897,
        -0.057500668,
        0.014008098,
        -0.06622493,
        -0.007072098,
        -0.09229578,
        0.03178172,
        -0.037777767,
        0.040862586,
        0.04502592,
        -0.009688233,
        0.10639735,
        0.016654536,
        -0.035804994,
        -0.03164334,
        0.043746058,
        0.00701212,
        -0.002278632,
        -0.034011792,
        -0.07318934,
        -0.6248428,
        0.048016872,
        -0.0051307557,
        -0.04352828,
        -0.14585584,
        0.06596451,
        -0.009511148,
        -0.012309287,
        0.05247607,
        -0.029942013,
        0.07302007,
        -0.035166226,
        -0.015321005,
        -0.0014265765,
        0.029512612,
        0.08514642,
        0.027568411
      ]
    },
    {
      "id": "worldsfair-speaker-88",
      "conference": "worldsfair",
      "name": "Aditya Gautam",
      "role": "Machine Learning Technical Lead",
      "company": "Meta",
      "twitter": "",
      "talks": [
        {
          "title": "Claude at Scale: Architecting the Context Stack for Large Codebases and Organizations",
          "description": "**Speaker Pitch** I'm an ML Lead at one of the largest technology companies in the world, building production AI systems across ML ranking, recommendation, and integrity at very large scale: codebases with hundreds of millions of lines, more than 10,000 engineers working in parallel, and models serving billions of users daily. I've given featured talks at Databricks Data + AI Summit 2025 (40,000+ attendees), AI Agent Conference NYC, MLOps Community, and Agent in Production '25, among others. My research on multi-agent LLM systems was published at ICWSM 2025 and I serve as a peer reviewer for NeurIPS, ICML, and AAAI. I have seen both sides of the full Claude ecosystem across every layer of the stack at this scale. I know what breaks: a stale CLAUDE.md propagating deprecated patterns silently, Skills that encode the wrong conventions and reproduce them reliably, Commands that work beautifully for one engineer and fall apart when shared across thousands. And I know what works: a hierarchical context stack with clear ownership boundaries at each layer, Skills and Commands designed as organizational contracts rather than personal shortcuts, Hooks and Plugins as the enforcement and auditability layer that makes the whole system trustworthy at scale. The mental model that ties it together, how to think about composing CLAUDE.md, Skills, Commands, Hooks, and Plugins as a coherent architecture rather than a collection of features, is what this talk is built around. This is a field report only I can give. --- **Session Description** At hyperscale, the question is never whether to use Claude. It is how to architect the Claude context stack so that thousands of engineers across infra, application, and ML research roles can best leverage it while remaining clear-eyed about its limitations and shortcomings. This talk moves past the basics and into the design decisions that determine whether your Claude deployment compounds productivity or compounds mistakes: how CLAUDE.md, Skills, Commands, Hooks, and Plugins compose into a coherent system, how each layer should be owned and optimized, and where the most expensive production failures actually originate. The mental model that unlocks this is treating the context stack as an architecture problem with the same discipline you would apply to any distributed system: clear ownership boundaries, explicit contracts between layers, and failure modes you have thought through before they hit production. The core of the talk is an honest account of both what breaks and what works, organized around a principled framework for how these components relate to each other and to the humans using them. CLAUDE.md encodes persistent architectural truth that must survive across every session and every engineer, and when it drifts it becomes the most dangerous file in your repository. Skills encode reusable procedural knowledge that, when designed as organizational contracts rather than personal shortcuts, become the highest-leverage investment a team can make in Claude adoption. Commands own the inner loop and work best when treated as shared infrastructure rather than individual automation. Hooks and Plugins are the enforcement and auditability layer, the part of the stack that makes the rest trustworthy at scale, and the part most teams underinvest in until something goes wrong. Getting these boundaries right looks completely different depending on the role: an infra platform team, an ML application team, and a research team each have a different optimal configuration, different risk surface, and different set of optimization opportunities. This talk maps all three explicitly. The practical payoff is a mental model and a set of design principles derived from both the failures and the wins at a company running some of the most complex ML systems in the world. You will leave knowing how to think about the Claude context stack as an architecture decision, how to optimize each layer for your team's specific role and risk profile, and what the anti-patterns are that look harmless at ten engineers and become critical failures at ten thousand.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        },
        {
          "title": "Modality Misalignment and Originality Attribution in Short-Form Video: A Multi-Agent Approach at Platform Scale",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.011235346,
        0.056827404,
        -0.02644633,
        0.06350621,
        -0.043350704,
        -0.041131627,
        -0.079881564,
        -0.033665456,
        0.045896996,
        -0.24717988,
        0.004825044,
        -0.04222815,
        0.025768496,
        0.00038200387,
        -0.04475478,
        0.021745415,
        0.08370069,
        -0.025472319,
        -0.07118296,
        -0.050597467,
        0.022080004,
        -0.05981532,
        -0.05299079,
        -0.0063338443,
        -0.008748679,
        0.04251779,
        -0.0243542,
        0.03782001,
        -0.14681946,
        0.46488756,
        -0.08731428,
        -0.047587037,
        -0.013842276,
        -0.022732984,
        0.09831192,
        -0.06656669,
        0.007949067,
        -0.0535391,
        0.013050169,
        0.017826889,
        0.00029013338,
        0.08225614,
        -0.042419948,
        0.055474874,
        -0.031186983,
        -0.015449447,
        -0.06193398,
        0.03763566,
        0.024417736,
        -0.075113356,
        -0.08735429,
        -0.023281703,
        0.061160628,
        -0.020773064,
        -0.03120323,
        0.03592331,
        0.033900544,
        -0.022787658,
        0.14501332,
        -0.0046298252,
        -0.09374353,
        0.004374178,
        -0.029731844,
        -0.027946645,
        -0.026345305,
        -0.0778997,
        0.008176544,
        -0.06762893,
        -0.047171034,
        0.02094781,
        0.0018804896,
        0.052232247,
        -0.062769756,
        -0.0059592244,
        -0.035281043,
        -0.1319412,
        -0.019721223,
        0.074738696,
        -0.010216994,
        0.023960082,
        -0.0029035998,
        0.036076643,
        -0.09796691,
        -0.087295406,
        0.07558038,
        -0.02350205,
        -0.10419545,
        0.014581632,
        -0.0680601,
        0.045804128,
        0.06413273,
        -0.096202634,
        -0.04010201,
        -0.041197628,
        -0.06480924,
        0.03210284,
        -0.013210532,
        0.04938521,
        0.043547567,
        0.051940925,
        0.06587956,
        -0.048856385,
        0.0061215404,
        -0.0054926854,
        -0.04625161,
        -0.008633219,
        0.019913575,
        -0.058387943,
        -0.0074025625,
        -0.038209442,
        -0.10816514,
        -0.5885569,
        0.03621889,
        0.074932784,
        -0.05937385,
        -0.14550316,
        0.030426843,
        -0.03592536,
        0.028245086,
        0.011642851,
        -0.046513468,
        0.04520024,
        -0.053478856,
        -0.0051867,
        -0.036246844,
        0.034537718,
        0.11850699,
        0.003315365
      ]
    },
    {
      "id": "worldsfair-speaker-89",
      "conference": "worldsfair",
      "name": "Arek Borucki",
      "role": "",
      "company": "Hugging Face",
      "twitter": "https://x.com/_Aras_B",
      "talks": [
        {
          "title": "Serving 2 Million Models Without Melting: Scaling the Hugging Face Hub",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0036737968,
        -0.01186428,
        0.03514248,
        0.031235691,
        0.031670157,
        -0.034778316,
        -0.09502193,
        -0.021202493,
        0.043418713,
        -0.17627051,
        -0.050075214,
        -0.06264795,
        0.061403558,
        -0.039694786,
        -0.090777,
        0.011779294,
        0.08350152,
        -0.052548144,
        -0.042251427,
        -0.054237578,
        -0.05670925,
        -0.08873636,
        0.017698744,
        0.06538511,
        -0.03562332,
        0.08811084,
        -0.027113276,
        0.011409417,
        -0.15263277,
        0.5001959,
        -0.086028405,
        -0.054698803,
        0.005314616,
        -0.014936775,
        0.01152817,
        -0.02531878,
        0.009749428,
        -0.010863096,
        -0.02158537,
        0.119116545,
        0.021680642,
        0.0064550913,
        -0.042381626,
        0.01257658,
        -0.045243107,
        0.0044328948,
        -0.04389437,
        0.02128213,
        -0.036769483,
        -0.03831499,
        -0.09144666,
        0.036653295,
        0.047889307,
        0.0066320812,
        -0.014574212,
        -0.0071525984,
        -0.016439784,
        -0.121578775,
        0.08950794,
        -0.09255182,
        -0.110256776,
        -0.060162652,
        0.04342001,
        0.029599188,
        -0.063064665,
        -0.07363645,
        0.018353973,
        -0.11244134,
        -0.025259297,
        -0.015375169,
        -0.06071706,
        -0.025024468,
        -0.0061806967,
        0.01397829,
        -0.04687759,
        0.020116756,
        0.052818898,
        0.08884428,
        -0.022170175,
        -0.07437879,
        -0.03319156,
        0.051944043,
        0.0087574795,
        -0.027958423,
        0.10062459,
        -0.03180852,
        -0.060230505,
        0.034859274,
        0.0059618116,
        0.1004131,
        0.10392928,
        0.034616392,
        -0.040809646,
        0.014164305,
        -0.017684292,
        -0.0073142853,
        -0.04924964,
        0.03232946,
        0.025325643,
        0.022804195,
        -0.040266227,
        -0.0026888999,
        0.024662275,
        -0.026887009,
        -0.04424522,
        -0.018892914,
        0.004414315,
        -0.03512333,
        0.037523706,
        -0.060011815,
        -0.04016331,
        -0.5919659,
        -0.019836828,
        -0.045721542,
        -0.05555747,
        -0.16725995,
        -0.06474464,
        0.024606453,
        -0.024185615,
        0.08119334,
        0.019436501,
        0.056940053,
        -0.034248456,
        0.03251221,
        0.055745192,
        0.04237236,
        0.05586858,
        0.033307515
      ]
    },
    {
      "id": "worldsfair-speaker-90",
      "conference": "worldsfair",
      "name": "Greg Pstrucha",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Stop prompting",
          "description": "In this talk I dive into usage of tooling, type systems and frameworks to enforce guardrails and limit slop produced by AI agents inside large codebases.",
          "day": "Day 2 — Session Day 1",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06599235,
        0.050149225,
        0.06498945,
        -0.030609045,
        0.031543806,
        0.003343864,
        -0.011252475,
        -0.008892583,
        0.060813837,
        -0.21815912,
        -0.03979936,
        0.018017992,
        -0.00051827217,
        -0.046388924,
        -0.017884556,
        0.041785713,
        0.069965824,
        -0.05819061,
        -0.055703998,
        -0.081623696,
        -0.07612906,
        -0.16863707,
        -0.015139769,
        0.08355564,
        -0.02228862,
        0.13323954,
        -0.009336481,
        0.05255824,
        -0.11904694,
        0.46110755,
        -0.066894665,
        0.029374821,
        0.03962309,
        -0.04132174,
        0.01826061,
        -0.03252776,
        -0.04981276,
        -0.027144559,
        0.0017099405,
        0.07820894,
        -0.061684094,
        0.013328469,
        -0.06492228,
        0.0045607807,
        0.004880006,
        -0.05363701,
        -0.10347695,
        -0.024197279,
        0.0075366334,
        -0.03897882,
        -0.041185617,
        0.018083312,
        -0.017020078,
        -0.008347996,
        -0.004128435,
        -0.004260276,
        -0.06471141,
        0.009494444,
        0.08492312,
        -0.058027875,
        -0.06783147,
        -0.02712576,
        -0.009121094,
        -0.022101905,
        -0.013389751,
        -0.033363633,
        0.06490813,
        -0.08066389,
        0.060482487,
        -0.067939945,
        -0.021119641,
        0.04162857,
        -0.08486978,
        0.03460591,
        0.0055790595,
        -0.093672045,
        -0.007970982,
        0.07457686,
        -0.044763595,
        0.02736734,
        -0.009410532,
        0.00573152,
        -0.049066212,
        -0.045564495,
        0.0661289,
        0.04405186,
        -0.026336757,
        0.013357392,
        -0.032243453,
        0.011888479,
        0.06757214,
        -0.05118568,
        -0.0495035,
        0.015726069,
        0.019208562,
        0.025228934,
        -0.0388531,
        0.016885111,
        0.035426833,
        0.021952268,
        -0.084009945,
        -0.06688203,
        -0.024589963,
        0.014929275,
        -0.02871863,
        0.0029598281,
        0.06024519,
        -0.043435648,
        0.008978725,
        0.013908468,
        0.0691372,
        -0.6183212,
        -0.035533633,
        0.044168945,
        -0.011941563,
        -0.16117896,
        -0.002468612,
        0.044712596,
        0.028291993,
        0.13304749,
        -0.038833376,
        0.07564156,
        -0.063838884,
        -0.05586057,
        0.016137421,
        0.08300013,
        0.035224915,
        0.035460025
      ]
    },
    {
      "id": "worldsfair-speaker-91",
      "conference": "worldsfair",
      "name": "Andrew Orobator",
      "role": "",
      "company": "Reddit",
      "twitter": "https://x.com/aorobator",
      "talks": [
        {
          "title": "Spin at the Gate Until Green: The Engineering Primitives Behind Self-Driving Codebases",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.026046151,
        0.050405744,
        0.055729732,
        0.01737649,
        0.017954512,
        0.01130802,
        -0.10232188,
        -0.005460631,
        0.08190013,
        -0.23233467,
        -0.02491343,
        -0.0750452,
        0.050519478,
        -0.027485289,
        0.0011849189,
        0.037673924,
        0.037673745,
        -0.059654627,
        0.00095582916,
        -0.023585845,
        -0.05680289,
        -0.088774644,
        0.055037625,
        0.018244121,
        -0.05849525,
        0.066763416,
        -0.043255154,
        0.06279,
        -0.10465224,
        0.4768036,
        -0.12176179,
        -0.010182229,
        -0.014294673,
        -0.02073077,
        0.07546061,
        -0.06154085,
        0.028603582,
        -0.023217265,
        0.01705784,
        0.043074507,
        -0.06432765,
        -0.088237084,
        -0.10382993,
        0.004782893,
        -0.02553189,
        0.013728641,
        -0.043446433,
        -0.032063812,
        -0.032271624,
        -0.049021486,
        -0.06686042,
        -0.0007008673,
        0.08405873,
        -0.04305249,
        0.024541784,
        0.045950305,
        -0.051681697,
        -0.055192426,
        0.10890136,
        -0.031981457,
        -0.0425567,
        -0.028612576,
        -0.0025157887,
        0.014783864,
        -0.017596217,
        -0.055865288,
        0.13594063,
        -0.049563844,
        0.027751932,
        0.019390903,
        -0.028863916,
        -0.013825847,
        -0.031900793,
        0.005842339,
        -0.02808502,
        -0.04291471,
        0.022551512,
        0.056385674,
        -0.04237661,
        0.0076342174,
        0.112360805,
        -0.040575374,
        0.0047119563,
        -0.01180345,
        -0.009624171,
        -0.049330823,
        -0.10371767,
        0.0058000907,
        0.04265859,
        0.01055352,
        0.023303512,
        -0.074518904,
        -0.015799109,
        -0.0024274318,
        -0.03885135,
        0.031231044,
        -0.12335175,
        0.09886759,
        0.03021448,
        0.054523457,
        -0.089943506,
        -0.04692985,
        0.07448452,
        -0.051538907,
        -0.0362323,
        -0.049674787,
        -0.06092707,
        -0.014676753,
        0.042602487,
        -0.0026506896,
        -0.011198553,
        -0.6042149,
        -0.0026588307,
        -0.04232499,
        0.015447509,
        -0.17744167,
        -0.012766463,
        0.03252455,
        -0.019486131,
        0.042457666,
        0.0029980023,
        0.050995264,
        -0.013011391,
        0.013319079,
        0.038349774,
        -0.020662483,
        0.064045705,
        0.021570543
      ]
    },
    {
      "id": "worldsfair-speaker-92",
      "conference": "worldsfair",
      "name": "David Hsu",
      "role": "",
      "company": "Retool",
      "twitter": "",
      "talks": [
        {
          "title": "Governance Is the Real Bottleneck to AI ROI",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.008852388,
        -0.010541852,
        -0.018099604,
        0.019261343,
        0.010410635,
        -0.021452706,
        -0.04762809,
        0.043680575,
        0.019622311,
        -0.21558098,
        -0.023821829,
        -0.069592714,
        0.006729953,
        0.033035573,
        -0.06688745,
        0.026698885,
        0.00071997836,
        -0.03772283,
        -0.06648142,
        -0.019798541,
        -0.08558029,
        -0.006403691,
        -0.0017818991,
        0.010147733,
        -0.031150449,
        0.056882124,
        0.050825085,
        0.0839794,
        -0.08613893,
        0.41737285,
        -0.083925016,
        0.008109421,
        0.050457187,
        0.028361779,
        0.03897057,
        -0.0619944,
        0.0080082435,
        -0.05671126,
        0.015789803,
        0.13957085,
        -0.014634392,
        -0.017304711,
        -0.046319876,
        0.042956144,
        -0.06780876,
        -0.03567451,
        -0.09939813,
        -0.044417962,
        -0.07933173,
        -0.019541344,
        -0.10923774,
        0.033003718,
        0.0011543166,
        -0.058829788,
        0.0337657,
        0.021512946,
        -0.05553503,
        -0.05130859,
        0.09038689,
        -0.009473571,
        -0.1855293,
        -0.013427413,
        -0.02639099,
        -0.06850774,
        0.024888316,
        -0.059274144,
        0.0500047,
        -0.06309084,
        -0.03363941,
        -0.01613135,
        -0.007644008,
        0.03337316,
        -0.0985476,
        -0.0047258595,
        -0.07090203,
        0.01429201,
        0.047499105,
        0.12308936,
        0.020166183,
        0.035840448,
        0.03998972,
        0.059233736,
        -0.12887812,
        -0.10155785,
        0.029757688,
        -0.023345416,
        -0.06569602,
        0.051320575,
        0.029819762,
        0.11178103,
        0.027156891,
        -0.087671876,
        -0.026384968,
        0.018399293,
        -0.12361404,
        0.038118012,
        -0.019542605,
        -0.0063255136,
        -0.04305389,
        0.024990382,
        -0.06481084,
        0.0035074889,
        -0.014010965,
        -0.012631011,
        -0.06631677,
        -0.014916873,
        0.013724987,
        -0.07016795,
        0.05012684,
        0.012331571,
        0.0004153637,
        -0.62725276,
        0.020790556,
        -0.024570728,
        0.028687721,
        -0.15219319,
        -0.04301374,
        0.026160786,
        0.053200763,
        0.05822904,
        0.00823032,
        0.071091086,
        -0.024207719,
        0.05283676,
        0.030250654,
        0.043062642,
        0.034422383,
        -0.006536882
      ]
    },
    {
      "id": "worldsfair-speaker-93",
      "conference": "worldsfair",
      "name": "Krishna Srinivasan",
      "role": "",
      "company": "Sarvam",
      "twitter": "https://x.com/fewshotlearner",
      "talks": [
        {
          "title": "From Scratch to SOTA: Training a 3B State-Space Vision Model for 1.4 Billion People",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.048836675,
        0.0376479,
        0.027895471,
        0.052833546,
        0.028289897,
        -0.017999291,
        -0.00008377947,
        -0.0055542886,
        0.07492498,
        -0.19638996,
        -0.04967711,
        -0.050164796,
        0.0029283576,
        -0.0464342,
        -0.042678732,
        -0.0018422424,
        0.023138259,
        -0.09455119,
        -0.07678352,
        -0.01242078,
        0.0036848683,
        -0.047978416,
        0.025906753,
        0.046001624,
        -0.06614323,
        0.053403713,
        -0.048060708,
        0.028119218,
        -0.12297995,
        0.4671418,
        -0.0745555,
        -0.051236212,
        -0.01755419,
        0.031787407,
        0.073206924,
        -0.087042525,
        0.003267735,
        -0.029551161,
        0.062595114,
        0.03724177,
        0.01674152,
        0.03512172,
        -0.027413256,
        0.044142112,
        -0.017045202,
        0.015922764,
        -0.0034367053,
        0.029623656,
        -0.028769549,
        -0.049643215,
        -0.09117146,
        -0.118597254,
        0.049214672,
        0.006134128,
        0.057741657,
        0.04530216,
        -0.018805189,
        -0.007525654,
        0.0456135,
        0.032972924,
        -0.15058215,
        0.022493666,
        0.05906897,
        0.010654656,
        -0.00932121,
        -0.039626226,
        0.053143293,
        -0.12259996,
        -0.05246516,
        0.057252314,
        0.033728376,
        -0.041694738,
        -0.07778482,
        0.072905704,
        -0.033793766,
        -0.059643142,
        -0.025923379,
        0.073173665,
        -0.023023618,
        0.0123507595,
        0.025493369,
        0.0477459,
        -0.05371832,
        -0.04280526,
        0.033466786,
        -0.048861537,
        -0.040947653,
        -0.010418141,
        0.0009086987,
        0.033384353,
        0.037260942,
        0.006205105,
        -0.02285388,
        -0.05184025,
        -0.047340915,
        0.06914304,
        -0.059738185,
        0.056034736,
        -0.0356271,
        -0.036388755,
        0.02483679,
        0.026603818,
        0.038475547,
        0.015862206,
        0.020829089,
        -0.02995381,
        0.09156214,
        -0.068806514,
        0.012354401,
        -0.031485192,
        -0.07778934,
        -0.6434698,
        -0.000086996166,
        0.00891363,
        -0.1010459,
        -0.1567721,
        -0.020818068,
        -0.045001075,
        0.08923405,
        0.03161702,
        0.000066352404,
        0.04699797,
        0.008932872,
        0.024032166,
        0.018537061,
        0.031585462,
        0.012153543,
        0.016045703
      ]
    },
    {
      "id": "worldsfair-speaker-94",
      "conference": "worldsfair",
      "name": "Mihnea Munteanu",
      "role": "Software Engineer",
      "company": "YouTube",
      "twitter": "https://x.com/MicneaPPK",
      "talks": [
        {
          "title": "Don't Summarize. Sample. — How YouTube Re-Built Search for the LLM Era",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04865475,
        -0.030861545,
        0.0040709344,
        -0.0047404417,
        0.04647498,
        -0.08176527,
        -0.048055444,
        0.046515375,
        0.058939204,
        -0.21308224,
        0.008214101,
        -0.03194339,
        0.04159277,
        -0.030867156,
        -0.059585817,
        0.082649335,
        0.07759512,
        -0.07965043,
        -0.08129863,
        -0.114592984,
        -0.055309974,
        -0.067721635,
        0.02838147,
        0.039448675,
        -0.06884865,
        0.059224732,
        0.0071137813,
        0.025721984,
        -0.1146207,
        0.51106995,
        -0.11486918,
        -0.056458175,
        -0.060673762,
        -0.011747187,
        0.043665875,
        -0.032284155,
        -0.023469329,
        0.016460594,
        0.011293339,
        0.079729654,
        -0.036100373,
        0.05243416,
        -0.03234241,
        0.0108287325,
        -0.039131552,
        -0.0451114,
        -0.049082648,
        -0.016995495,
        -0.03722419,
        0.032932013,
        -0.06883199,
        0.0007154385,
        0.013858809,
        -0.017989656,
        -0.0023373533,
        0.021954183,
        0.0026158888,
        -0.052181005,
        0.09017043,
        -0.01651332,
        -0.09822675,
        0.01861702,
        0.00045470177,
        0.026904035,
        0.0022825445,
        -0.06775227,
        0.0427464,
        -0.12231211,
        -0.0027709024,
        0.0015574498,
        0.0024077168,
        -0.021237852,
        -0.02253887,
        -0.04756309,
        0.026258688,
        -0.0735468,
        0.003931546,
        0.08390461,
        -0.008123678,
        -0.022038316,
        0.032131743,
        -0.010265003,
        -0.05587055,
        0.0008545017,
        0.06048369,
        -0.019037046,
        -0.081570014,
        0.010391598,
        -0.023930242,
        0.07126161,
        -0.049301777,
        -0.03290988,
        -0.023301324,
        -0.009088402,
        -0.03492147,
        0.009887854,
        -0.06270954,
        -0.015225665,
        0.04027757,
        -0.013026065,
        -0.015755774,
        -0.015538943,
        0.024910001,
        -0.023513306,
        -0.006451023,
        -0.03993865,
        0.019546457,
        -0.041404072,
        0.08894265,
        -0.09272674,
        -0.068332694,
        -0.6157255,
        -0.033104654,
        0.0020169762,
        -0.03341281,
        -0.16663048,
        0.02766291,
        -0.006370136,
        0.040422235,
        0.042755127,
        0.017812978,
        0.020161504,
        -0.058833916,
        -0.024448141,
        0.060318116,
        0.021851055,
        0.05851012,
        0.01385222
      ]
    },
    {
      "id": "worldsfair-speaker-95",
      "conference": "worldsfair",
      "name": "Eugene Yan",
      "role": "Member of Technical Staff",
      "company": "Anthropic",
      "twitter": "https://x.com/eugeneyan",
      "talks": [
        {
          "title": "Using LLMs to Secure Source Code",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.04049944,
        -0.017476209,
        -0.019401394,
        0.010344823,
        0.051730692,
        0.0027625342,
        -0.06412177,
        -0.03325669,
        0.022738406,
        -0.20408495,
        -0.012793308,
        -0.09212578,
        -0.0024500377,
        0.008634599,
        -0.05060664,
        0.034892928,
        0.034786724,
        -0.082408495,
        -0.105066694,
        -0.09030933,
        -0.028817035,
        -0.0965418,
        0.08952781,
        0.0551191,
        -0.042032532,
        0.078504786,
        -0.02153407,
        0.073768124,
        -0.13504863,
        0.4410266,
        -0.040767465,
        0.016880022,
        0.043952532,
        -0.062419172,
        0.04009933,
        -0.073271096,
        0.0065048244,
        -0.015278037,
        0.030791169,
        0.0027462605,
        -0.02899614,
        -0.018770708,
        -0.04944345,
        0.053646445,
        -0.06816733,
        0.029584272,
        -0.1129055,
        -0.027433923,
        -0.013682174,
        -0.014360551,
        -0.087016776,
        -0.0065183653,
        0.060731456,
        -0.059075978,
        0.009078944,
        0.016764855,
        0.027316265,
        -0.07008282,
        0.09779844,
        -0.0035867402,
        -0.09657098,
        -0.033668507,
        0.007334556,
        0.045238197,
        -0.03183066,
        -0.10379382,
        0.08777345,
        -0.07535522,
        0.057563383,
        -0.0068010865,
        0.0069378954,
        0.048489045,
        -0.061391048,
        0.011960841,
        -0.10579412,
        -0.036424715,
        0.024018506,
        0.06816181,
        -0.061832387,
        -0.05109358,
        0.06404876,
        0.097977825,
        -0.051487483,
        -0.07419836,
        0.053426117,
        -0.03288553,
        -0.003014878,
        0.058915965,
        -0.024724128,
        0.03252401,
        0.03209389,
        -0.065045,
        -0.046779968,
        0.010777011,
        -0.11600451,
        0.06428136,
        -0.06890485,
        0.015350769,
        -0.03530913,
        0.008622085,
        0.004764134,
        -0.03606884,
        0.058087483,
        -0.037360188,
        -0.08498725,
        -0.07455662,
        -0.020713132,
        -0.010935043,
        0.064793356,
        -0.0057973494,
        -0.09041817,
        -0.59521335,
        -0.00007346372,
        -0.00682142,
        -0.051392168,
        -0.15886527,
        0.0024707953,
        -0.07401066,
        0.08130739,
        0.08498236,
        -0.084698565,
        0.052597377,
        -0.0398905,
        -0.0062639294,
        -0.016172927,
        0.066351324,
        0.061630853,
        -0.018577125
      ]
    },
    {
      "id": "worldsfair-speaker-96",
      "conference": "worldsfair",
      "name": "Vyas A",
      "role": "",
      "company": "Plivo",
      "twitter": "https://x.com/not_ryan_vy45",
      "talks": [
        {
          "title": "5 Voice Agent Failure Modes You'll Hit in Week One",
          "description": "A practical talk on the five voice-agent failures teams hit immediately in production: interruptions, turn-taking misfires, compounding latency, hallucinated actions, and audio/transcription mismatches. Each failure comes with a real example and a concrete fix.",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.030528365,
        0.04086536,
        0.080013484,
        0.032450486,
        0.04640992,
        0.01854161,
        -0.0715348,
        0.051589143,
        0.02604356,
        -0.22272284,
        0.0060785357,
        -0.07049524,
        0.005113616,
        -0.09603398,
        -0.034042064,
        0.016983958,
        0.020842906,
        -0.09404818,
        -0.07283647,
        -0.0551286,
        0.0026944526,
        -0.12550373,
        -0.07305881,
        0.060337998,
        -0.09549989,
        0.036022156,
        -0.038249806,
        0.06273238,
        -0.07708537,
        0.45492378,
        -0.100751735,
        -0.038607758,
        0.041374106,
        -0.052192613,
        0.021695849,
        -0.06734254,
        0.02308294,
        -0.005187329,
        0.0100936275,
        0.03353179,
        0.04770549,
        0.07622171,
        -0.030198136,
        0.011693904,
        0.028757496,
        0.001956447,
        -0.058601037,
        0.017291462,
        -0.0785438,
        -0.03620279,
        -0.064047076,
        -0.023199199,
        -0.03759346,
        0.014206462,
        0.0102254525,
        0.05178227,
        -0.03333464,
        -0.065990396,
        0.109854996,
        -0.042054944,
        -0.054604113,
        0.0101032,
        0.036763832,
        -0.042280518,
        -0.035607416,
        -0.06134965,
        0.014207527,
        -0.11201614,
        -0.031058045,
        -0.0089533795,
        -0.011668773,
        0.010049107,
        -0.05238178,
        0.050943624,
        -0.030361256,
        -0.11686737,
        -0.033080142,
        0.046716083,
        0.013338133,
        -0.0012040994,
        -0.0034708013,
        0.079043925,
        -0.017411927,
        -0.039361626,
        -0.020134017,
        -0.033448156,
        -0.022680892,
        0.0017557263,
        -0.047902573,
        0.08888034,
        0.024251359,
        -0.062519446,
        -0.058056477,
        -0.000012862241,
        -0.06677617,
        0.00602788,
        -0.0067394013,
        -0.014342451,
        -0.06824527,
        0.0064363447,
        -0.011297794,
        0.0213326,
        0.03736208,
        0.042477176,
        -0.02064634,
        -0.049749468,
        -0.031139527,
        0.009991891,
        -0.013350337,
        -0.0492432,
        -0.05017391,
        -0.6461637,
        -0.013965186,
        -0.0702606,
        -0.027180161,
        -0.1648386,
        -0.011275182,
        0.031815477,
        -0.016782105,
        0.07869275,
        0.037607115,
        0.06489547,
        -0.0065514804,
        -0.017480938,
        -0.03849772,
        0.049708717,
        0.052438125,
        -0.012572501
      ]
    },
    {
      "id": "worldsfair-speaker-97",
      "conference": "worldsfair",
      "name": "Nachiket Paranjape",
      "role": "",
      "company": "DoorDash",
      "twitter": "https://x.com/nmparanjape",
      "talks": [
        {
          "title": "AI Evals Platform for Cross-Functional Teams at Scale",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.046789765,
        0.08193941,
        -0.001780057,
        0.097372904,
        -0.008290542,
        -0.0039086775,
        -0.08912604,
        0.035790708,
        0.077633366,
        -0.1976455,
        -0.036152698,
        -0.011556809,
        0.022191763,
        0.03273575,
        -0.031875644,
        0.010621205,
        0.058720272,
        -0.056614798,
        -0.05304887,
        -0.083651796,
        -0.080066994,
        -0.02038094,
        0.023827277,
        0.03599313,
        -0.051588386,
        0.05152742,
        0.07426625,
        0.058138143,
        -0.12054125,
        0.49958426,
        -0.10625814,
        -0.06076404,
        0.043237668,
        -0.03378214,
        0.05730682,
        -0.014375376,
        0.018351771,
        0.0038684334,
        0.023273908,
        0.062379364,
        -0.013880055,
        0.067560874,
        -0.0642377,
        0.025252622,
        -0.09367584,
        -0.00925139,
        -0.06895848,
        -0.030852327,
        -0.005062701,
        -0.046845827,
        -0.08149222,
        -0.07237718,
        -0.001556778,
        -0.06890557,
        0.032851014,
        0.04234569,
        -0.012980901,
        -0.04102432,
        0.12705737,
        -0.07606785,
        -0.10767375,
        -0.013760343,
        0.0042791055,
        -0.03630294,
        0.008186114,
        -0.061681196,
        0.055277076,
        -0.06154881,
        0.0021733437,
        -0.006625588,
        0.068703756,
        0.0026144553,
        -0.02738178,
        0.07808194,
        -0.041479774,
        -0.040185165,
        -0.0081639225,
        0.03448067,
        -0.04068148,
        -0.012032454,
        0.08008418,
        0.07147058,
        -0.035024885,
        -0.022434633,
        0.013799922,
        0.0014601396,
        -0.054033052,
        -0.022615688,
        -0.025889887,
        0.050646164,
        0.02825041,
        -0.050752074,
        0.0086132875,
        -0.024646644,
        0.0027586522,
        0.018852971,
        -0.09284768,
        0.01145122,
        -0.0068519195,
        0.02741857,
        0.004689169,
        -0.0073342454,
        0.026471172,
        0.084975004,
        -0.020390127,
        -0.025780827,
        0.013001481,
        -0.04924698,
        0.022255769,
        -0.031756923,
        -0.02670007,
        -0.60161555,
        0.0018518246,
        -0.01023755,
        -0.053966664,
        -0.15510796,
        0.05135849,
        0.0103288125,
        0.090241574,
        0.12524138,
        0.065542206,
        0.030420909,
        -0.09064973,
        -0.037035625,
        -0.013665375,
        -0.009980593,
        0.0497671,
        -0.027444677
      ]
    },
    {
      "id": "worldsfair-speaker-98",
      "conference": "worldsfair",
      "name": "Swaroop Chitlur Haridas",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "AI Evals Platform for Cross-Functional Teams at Scale",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.086195216,
        0.10920686,
        -0.011995583,
        0.10628299,
        0.0025946463,
        -0.04049114,
        -0.078949265,
        -0.04076731,
        0.053815868,
        -0.20798543,
        0.006982911,
        -0.022169026,
        0.041375,
        -0.054472532,
        0.029430782,
        0.007674551,
        0.064035594,
        -0.020713145,
        -0.060795166,
        -0.06970795,
        -0.061087716,
        -0.007933803,
        -0.00960941,
        0.05745387,
        -0.042923763,
        0.08116577,
        0.07297669,
        0.08219028,
        -0.13922763,
        0.50581014,
        -0.082007214,
        -0.060107607,
        0.030123234,
        -0.012091051,
        0.047430504,
        -0.0074868184,
        -0.01897414,
        0.004459649,
        0.058404565,
        0.073833,
        0.039869875,
        0.032754987,
        -0.04262424,
        0.04684576,
        -0.028299209,
        -0.025457801,
        -0.04964558,
        -0.042514518,
        -0.06624099,
        -0.03301176,
        -0.06275929,
        -0.115903154,
        0.07117103,
        0.011905445,
        -0.003983926,
        0.06811619,
        -0.028572854,
        -0.039150506,
        0.07903904,
        -0.0076138014,
        -0.07459359,
        0.007048634,
        0.043643598,
        -0.067022696,
        0.0565669,
        -0.04897353,
        0.0828615,
        -0.10704576,
        0.0027681247,
        -0.022652617,
        0.04724166,
        -0.005299401,
        0.0092079695,
        0.01359225,
        -0.031744245,
        -0.041353974,
        0.006399433,
        0.055125695,
        -0.015271224,
        -0.039391823,
        0.051347133,
        0.059770476,
        0.0133390585,
        -0.048529133,
        0.0028726936,
        -0.00042586782,
        -0.046127036,
        0.03178757,
        0.0028536175,
        0.0045289183,
        0.037990548,
        -0.033920255,
        -0.030933827,
        -0.013648072,
        -0.022703264,
        0.029217456,
        -0.12717466,
        0.018943937,
        -0.091124155,
        0.0025938572,
        -0.047425125,
        -0.017669654,
        0.007046003,
        0.03607293,
        -0.009442869,
        -0.03129651,
        0.049127027,
        -0.035325114,
        0.0061713643,
        -0.015896844,
        -0.026232544,
        -0.60503894,
        -0.001604754,
        -0.005083079,
        -0.050068125,
        -0.14649618,
        -0.019058257,
        0.0045938077,
        0.060235985,
        0.09512941,
        0.011003018,
        0.05076212,
        -0.0778631,
        0.016497353,
        -0.009127272,
        0.040644594,
        0.046215646,
        -0.023710461
      ]
    },
    {
      "id": "worldsfair-speaker-99",
      "conference": "worldsfair",
      "name": "Aman Raj",
      "role": "",
      "company": "Decawork",
      "twitter": "",
      "talks": [
        {
          "title": "IT Admin for the AI Workforce: Why Your AI Agents Will Need Their Own IT Department",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.040202368,
        0.09162216,
        -0.0060752,
        -0.0057868566,
        0.03836306,
        -0.05532661,
        -0.055568278,
        0.058894213,
        -0.007401556,
        -0.238969,
        0.014288167,
        -0.073510356,
        0.013298246,
        0.05089156,
        -0.043398555,
        -0.011126225,
        0.08347173,
        -0.05890581,
        -0.05547876,
        -0.07707846,
        -0.044498224,
        -0.045404214,
        -0.007060804,
        0.027459728,
        -0.08517693,
        0.10511621,
        0.029525891,
        0.037907027,
        -0.109043844,
        0.4672338,
        -0.038174752,
        -0.015478507,
        0.09548654,
        -0.03407223,
        0.030785594,
        -0.047670815,
        -0.0112239905,
        0.036568623,
        0.0051802085,
        0.066711135,
        -0.06297069,
        0.048276633,
        -0.09570609,
        -0.00052570587,
        -0.025443252,
        0.05846413,
        -0.05531804,
        -0.016555201,
        0.040201087,
        -0.08006762,
        -0.033650614,
        0.05149638,
        0.03643682,
        -0.016480714,
        -0.029472772,
        0.03214588,
        -0.03186555,
        -0.11501932,
        0.078525946,
        -0.02594164,
        -0.12922312,
        -0.06689741,
        -0.011093523,
        0.023386896,
        -0.064266965,
        -0.069295906,
        0.071260914,
        -0.06448014,
        -0.03678908,
        -0.036767475,
        0.029050326,
        0.032844376,
        0.0032020386,
        0.018677924,
        -0.075985834,
        -0.036425356,
        -0.0010447851,
        0.11144274,
        -0.026801433,
        0.025510255,
        -0.0079018315,
        0.018669004,
        -0.11110231,
        -0.05739436,
        0.04912408,
        -0.027294733,
        -0.039046805,
        0.065660566,
        0.014527805,
        0.056793023,
        0.038104348,
        -0.009108097,
        0.014327742,
        0.019927232,
        -0.05347654,
        0.037253764,
        -0.031863604,
        0.07609577,
        0.0122128185,
        0.017273264,
        0.011693655,
        -0.024367554,
        0.033997703,
        -0.000043727556,
        0.005515947,
        -0.053437024,
        0.052102823,
        -0.018298227,
        0.030205525,
        0.0033143933,
        -0.0059386804,
        -0.6172786,
        -0.008194665,
        0.05209743,
        -0.058462594,
        -0.16039814,
        -0.010337222,
        -0.00502061,
        0.025550526,
        0.042138483,
        -0.049476266,
        0.096150376,
        0.0016878161,
        -0.0069403634,
        0.018741265,
        0.09861711,
        -0.0053460044,
        -0.040361322
      ]
    },
    {
      "id": "worldsfair-speaker-100",
      "conference": "worldsfair",
      "name": "Daksh Gupta",
      "role": "",
      "company": "Greptile",
      "twitter": "https://x.com/greptile",
      "talks": [
        {
          "title": "What we learned by analyzing 1M AI-generated PRs",
          "description": "Charlie Holtz (@charlieholtz) - Founder & CEO of Conductor (launched Jul 2025). Mac app for orchestrating multiple coding agents in parallel. Used at YC cos, Linear, Vercel, Notion, Supabase. Ex-Replicate, ex-Point72, Brown CS. Angle: parallel agent orchestration / humans-as-conductors - distinct from the rest of the Coding Agents track. Ref tweet: https://x.com/charlieholtz/status/2047351098634338610",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.017592654,
        0.043755375,
        0.05135712,
        0.020765418,
        0.013988015,
        -0.019501196,
        -0.0927001,
        0.050020497,
        0.022966547,
        -0.25205976,
        -0.07372193,
        -0.04780011,
        -0.031236771,
        -0.032404102,
        -0.044732742,
        -0.030956108,
        0.07039194,
        -0.06593582,
        -0.101833515,
        -0.033481114,
        -0.056941163,
        -0.050791226,
        -0.06227786,
        0.057477333,
        -0.017014816,
        0.1011521,
        0.0031999224,
        0.07930461,
        -0.11630463,
        0.4604656,
        -0.100413516,
        -0.038800903,
        0.05006113,
        -0.0041005025,
        0.03346795,
        -0.07113232,
        0.020326814,
        0.011414428,
        0.019216152,
        0.06356933,
        -0.045246907,
        0.029241467,
        -0.060187504,
        0.03398727,
        -0.036255628,
        -0.036291808,
        -0.0665377,
        -0.011644828,
        -0.014849218,
        -0.024862574,
        -0.043630775,
        -0.0074702417,
        -0.0047320696,
        0.022944072,
        0.022322383,
        0.044574454,
        -0.05394461,
        -0.06900826,
        0.044794995,
        0.000237669,
        -0.08880198,
        -0.0367352,
        0.07835799,
        0.0015679309,
        0.0776876,
        -0.03602545,
        0.013292207,
        -0.058166333,
        0.009819311,
        -0.08353909,
        0.02510119,
        0.01806456,
        -0.083275594,
        0.009493093,
        -0.029898956,
        -0.10705325,
        -0.0010960472,
        0.08050871,
        0.010277168,
        0.019162148,
        -0.048553128,
        0.079918556,
        -0.10974134,
        -0.061523207,
        0.02350844,
        0.029378463,
        -0.0042730686,
        0.0060132938,
        0.009238765,
        0.03311412,
        0.04035558,
        -0.036238957,
        -0.024071408,
        0.009057833,
        -0.09744389,
        0.040268492,
        -0.047179446,
        0.015244109,
        0.016483331,
        0.045496687,
        0.043284304,
        -0.027541986,
        0.0395368,
        0.016770924,
        -0.06626129,
        -0.06200039,
        0.04217897,
        0.010022524,
        -0.07213685,
        -0.008715736,
        -0.0274439,
        -0.61425674,
        -0.0016342207,
        0.08671241,
        -0.04100652,
        -0.15096746,
        0.0038319586,
        0.0014225941,
        0.06853824,
        0.113509424,
        -0.0015484877,
        0.056507833,
        -0.056337442,
        0.00017160378,
        -0.02463929,
        0.009875523,
        0.111702666,
        -0.001858713
      ]
    },
    {
      "id": "worldsfair-speaker-101",
      "conference": "worldsfair",
      "name": "Karan Vaidya",
      "role": "",
      "company": "Composio",
      "twitter": "https://x.com/KaranVaidya6",
      "talks": [
        {
          "title": "Tool Execution layer for agents",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0599651,
        0.076235935,
        0.014784898,
        0.06682552,
        -0.007900786,
        0.0026384306,
        -0.07552015,
        0.046838444,
        -0.027827887,
        -0.20343053,
        -0.049046747,
        -0.06914144,
        0.03385169,
        -0.02749311,
        -0.016382625,
        -0.005011909,
        0.08938334,
        -0.10166682,
        -0.061013967,
        -0.062393293,
        -0.03272382,
        -0.0069830455,
        -0.05430536,
        0.0075107804,
        -0.035411723,
        0.059887476,
        -0.04410154,
        0.012496485,
        -0.08962214,
        0.49381766,
        -0.036538396,
        -0.028328186,
        0.016789753,
        -0.0024411434,
        0.0017965055,
        -0.08081187,
        -0.014212612,
        0.012059681,
        0.002586995,
        0.058031995,
        -0.06008796,
        0.050057296,
        -0.03757019,
        0.03375321,
        0.026660701,
        0.034068145,
        -0.035081238,
        0.015860721,
        -0.04831634,
        -0.051555328,
        -0.07700299,
        0.011512344,
        0.047823384,
        -0.054079317,
        0.03357972,
        -0.003532622,
        -0.040023204,
        -0.09689437,
        0.083208844,
        -0.018826257,
        -0.071488984,
        -0.041686647,
        0.057355177,
        -0.012689493,
        -0.021762613,
        -0.061934177,
        0.046255115,
        -0.101979494,
        0.013443029,
        -0.019393153,
        0.0631508,
        -0.008202106,
        -0.037711248,
        0.07330467,
        -0.06021824,
        -0.07467324,
        -0.008456314,
        0.13893665,
        0.048807424,
        0.00036900587,
        0.018134162,
        -0.026055971,
        -0.071208656,
        -0.12804817,
        -0.037431654,
        0.0023757468,
        -0.03697832,
        0.04427278,
        -0.02595538,
        0.05731814,
        0.07067697,
        -0.041378878,
        -0.037121862,
        -0.035351392,
        -0.023718903,
        -0.0040027145,
        -0.05066873,
        0.0013888548,
        -0.057688184,
        0.027086385,
        0.018435985,
        -0.065202646,
        0.04327226,
        0.040415574,
        -0.08219397,
        -0.0378734,
        0.07034626,
        -0.010629916,
        0.015030812,
        -0.08031494,
        -0.028030835,
        -0.6179023,
        -0.020207085,
        -0.04708121,
        -0.00922646,
        -0.15129586,
        0.0029019495,
        0.0125143165,
        0.03584001,
        0.11733524,
        -0.012523401,
        0.058388796,
        0.003242949,
        -0.027138911,
        -0.030133726,
        -0.022571512,
        -0.0046605775,
        -0.05212488
      ]
    },
    {
      "id": "worldsfair-speaker-102",
      "conference": "worldsfair",
      "name": "Cody Menefee",
      "role": "",
      "company": "Firecrawl",
      "twitter": "https://x.com/cbmenefee",
      "talks": [
        {
          "title": "You’re Not Thinking Big Enough: Rebuilding Food Systems from First Principles with AI Agents",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.015645111,
        0.014601806,
        0.033933878,
        -0.0021827288,
        0.022017488,
        0.045601323,
        -0.05128823,
        -0.043071598,
        0.0016205991,
        -0.19465083,
        0.0012557076,
        -0.07847219,
        -0.013384864,
        0.019240618,
        -0.083901085,
        0.05635757,
        0.057159733,
        -0.083062164,
        -0.02786153,
        -0.06660171,
        -0.11057563,
        -0.121675946,
        -0.027677394,
        0.037630264,
        -0.029287813,
        0.08070081,
        -0.010693673,
        0.06487426,
        -0.08932138,
        0.4929364,
        -0.08484349,
        0.01719834,
        0.050612383,
        0.056578483,
        0.08077843,
        -0.074888356,
        -0.08691706,
        0.03805613,
        0.022362072,
        0.034125928,
        -0.038227864,
        0.009639141,
        0.000055545846,
        0.0014347198,
        -0.051352285,
        0.07547375,
        -0.06150759,
        -0.054540973,
        0.041037433,
        -0.05076724,
        -0.082413584,
        0.035950866,
        0.06477211,
        -0.020318108,
        0.05520311,
        0.029003695,
        0.020838153,
        -0.061777048,
        0.014413116,
        0.019491155,
        -0.020331636,
        -0.0017552701,
        -0.07844078,
        0.08332443,
        -0.0017639634,
        -0.08942058,
        0.06603775,
        -0.0425562,
        0.021349698,
        -0.04070082,
        0.040191088,
        0.07363922,
        -0.009217367,
        -0.012247301,
        -0.089479774,
        -0.039934482,
        0.0045439615,
        0.07408512,
        -0.0245295,
        -0.006750594,
        -0.008921437,
        0.0002487618,
        -0.030705217,
        -0.03748412,
        -0.008754943,
        -0.11782501,
        -0.02793372,
        0.051874947,
        -0.016675564,
        -0.0182621,
        0.008818465,
        -0.04107395,
        0.017348567,
        -0.014529751,
        -0.05089224,
        -0.01708181,
        -0.0632567,
        -0.021319147,
        0.0075941556,
        -0.000020764382,
        -0.061851375,
        -0.0869173,
        0.05877914,
        -0.02061418,
        -0.07232239,
        -0.010243098,
        0.084534764,
        0.019390438,
        -0.025045767,
        -0.03677226,
        -0.037271336,
        -0.6156279,
        -0.08025616,
        0.001674372,
        0.00083370565,
        -0.16183305,
        0.011827137,
        0.05270141,
        0.034746498,
        0.033662252,
        -0.041247107,
        0.00927078,
        -0.0059439857,
        0.011027499,
        0.06772272,
        -0.020062465,
        0.096941195,
        0.056628887
      ]
    },
    {
      "id": "worldsfair-speaker-103",
      "conference": "worldsfair",
      "name": "Hiral Shah",
      "role": "Senior Director of Product",
      "company": "Docusign",
      "twitter": "",
      "talks": [
        {
          "title": "Your Agreements Are a Database You Can't Query. We're Fixing That",
          "description": "Agreements power every enterprise business, but mismanaging them has a big cost. Deloitte and Docusign found that inefficient agreement management costs businesses nearly $2 trillion in lost economic value annually, and organizations burn more than 55 billion hours on manual, disconnected agreement workflows. This session walks through what it takes to turn agreement repositories into structured, queryable data that can power search, analytics, and agentic AI workflows using Nvidia's open AI platform as the foundation. We'll cover the full pipeline: Extraction: How Nemotron Parse handles the hardest document understanding problems, including contract tables where pricing tiers, SLAs, and payment schedules live in dense layouts that traditional OCR destroys. Embedding and retrieval: How NeMo Retriever's embedding and reranking models make extracted contract data searchable and surfaceable for RAG-based applications. Agentic workflows: How these components connect to power contract analysis agents that can answer questions, surface risks, and act on obligations at scale. Nvidia will present the platform architecture and model capabilities. Docusign will share what we've learned evaluating this stack against real customer agreements: what worked, what broke, and what production deployment actually requires when serving over a billion users. Attendees will leave with a realistic view of where the Nvidia AI platform excels at document intelligence, where the gaps remain, and how to think about building agentic contract workflows in their own organizations.",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.014467282,
        0.07980148,
        -0.03494384,
        0.0019464729,
        0.013489214,
        -0.0371531,
        -0.108975954,
        0.029226366,
        0.020767732,
        -0.17822272,
        0.06533523,
        -0.060603127,
        -0.025840452,
        -0.0302643,
        -0.08926376,
        -0.04354282,
        0.073673055,
        0.0033412166,
        -0.1030628,
        -0.09647769,
        -0.063705906,
        -0.037204966,
        -0.03529622,
        0.10845271,
        -0.11142042,
        0.07097381,
        0.081881106,
        0.08368683,
        -0.13562629,
        0.45923734,
        -0.075288445,
        -0.038834162,
        0.0109293265,
        -0.06763317,
        0.040365063,
        -0.045360625,
        -0.05098443,
        0.008744339,
        -0.011866551,
        0.045480967,
        0.0468932,
        0.027153507,
        -0.043094456,
        0.03992955,
        -0.08109036,
        -0.06548538,
        -0.052074324,
        0.05040198,
        -0.045515664,
        -0.040945724,
        -0.06935681,
        -0.009637207,
        0.043743446,
        -0.040425498,
        0.008235376,
        0.014141732,
        -0.0099365385,
        -0.039301056,
        0.07319493,
        -0.0692486,
        -0.13392325,
        0.040585846,
        -0.061542004,
        -0.011252793,
        0.03318886,
        -0.083868906,
        0.04358276,
        0.027538426,
        -0.017661503,
        0.009635519,
        -0.02783863,
        0.043361966,
        -0.05839453,
        0.028052352,
        -0.066477284,
        -0.15287437,
        -0.04287904,
        0.10618436,
        -0.006431793,
        0.009130849,
        -0.0008308324,
        0.05461156,
        -0.043972917,
        -0.07349881,
        0.07612276,
        -0.05649796,
        -0.086021416,
        0.04254428,
        -0.06333499,
        0.08584919,
        -0.018672345,
        -0.067676194,
        -0.004944531,
        0.016143572,
        -0.056032345,
        0.035324484,
        -0.042915564,
        0.05439943,
        -0.056170966,
        0.06626926,
        0.028685037,
        -0.07475835,
        0.019060537,
        0.062271457,
        0.015865363,
        -0.00020627865,
        0.010719017,
        -0.035338517,
        0.09801494,
        -0.04314741,
        -0.03126622,
        -0.57601815,
        0.043828145,
        -0.014650305,
        -0.005525196,
        -0.1390843,
        0.036311734,
        -0.07241588,
        0.08004793,
        0.039197877,
        -0.024874039,
        0.016143722,
        -0.05732047,
        -0.034183417,
        -0.06380856,
        0.01262605,
        -0.012578243,
        0.09037618
      ]
    },
    {
      "id": "worldsfair-speaker-104",
      "conference": "worldsfair",
      "name": "Sean Sodha",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Your Agreements Are a Database You Can't Query. We're Fixing That",
          "description": "Agreements power every enterprise business, but mismanaging them has a big cost. Deloitte and Docusign found that inefficient agreement management costs businesses nearly $2 trillion in lost economic value annually, and organizations burn more than 55 billion hours on manual, disconnected agreement workflows. This session walks through what it takes to turn agreement repositories into structured, queryable data that can power search, analytics, and agentic AI workflows using Nvidia's open AI platform as the foundation. We'll cover the full pipeline: Extraction: How Nemotron Parse handles the hardest document understanding problems, including contract tables where pricing tiers, SLAs, and payment schedules live in dense layouts that traditional OCR destroys. Embedding and retrieval: How NeMo Retriever's embedding and reranking models make extracted contract data searchable and surfaceable for RAG-based applications. Agentic workflows: How these components connect to power contract analysis agents that can answer questions, surface risks, and act on obligations at scale. Nvidia will present the platform architecture and model capabilities. Docusign will share what we've learned evaluating this stack against real customer agreements: what worked, what broke, and what production deployment actually requires when serving over a billion users. Attendees will leave with a realistic view of where the Nvidia AI platform excels at document intelligence, where the gaps remain, and how to think about building agentic contract workflows in their own organizations.",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.041821267,
        0.09747486,
        -0.0010339046,
        0.050873745,
        0.059544753,
        0.026530115,
        -0.097720146,
        0.07586789,
        0.03118893,
        -0.22446477,
        0.013146072,
        -0.048884623,
        0.0033491394,
        -0.028354501,
        -0.036784098,
        -0.025918389,
        0.055457905,
        -0.0012903479,
        -0.090433605,
        -0.098269895,
        -0.025162263,
        -0.13586324,
        -0.04586014,
        0.043816306,
        -0.11631272,
        0.0513725,
        -0.015355027,
        0.07326882,
        -0.13868426,
        0.47559753,
        -0.01558861,
        -0.033167314,
        -0.025765475,
        -0.056063093,
        0.027750194,
        -0.029012106,
        -0.063916095,
        0.056262333,
        0.037585206,
        0.07079339,
        -0.012438202,
        0.026170641,
        -0.05416989,
        0.04455347,
        -0.041825864,
        -0.058708537,
        -0.06916505,
        0.045918453,
        -0.051188104,
        -0.019910926,
        -0.049772557,
        0.0030663395,
        0.06739059,
        -0.06771085,
        -0.013277218,
        0.042987414,
        0.016860038,
        -0.016974535,
        0.07575173,
        -0.08615006,
        -0.16760781,
        -0.009581047,
        -0.061948895,
        0.018932458,
        -0.040584836,
        -0.04456062,
        0.056000777,
        -0.002654087,
        -0.011942964,
        -0.058213092,
        -0.05076638,
        0.038363427,
        -0.032249928,
        -0.0023442665,
        -0.013507113,
        -0.13090727,
        -0.029526738,
        0.07950782,
        -0.0017383259,
        0.03433548,
        -0.017606478,
        0.039655674,
        -0.05571887,
        -0.055352274,
        0.043323085,
        -0.022163352,
        -0.07700939,
        0.08914678,
        -0.022892892,
        0.07952221,
        0.022485068,
        -0.010729834,
        -0.021923743,
        -0.004340085,
        -0.045047656,
        0.022682516,
        -0.05149521,
        0.06300687,
        -0.08486634,
        -0.058343768,
        0.07605161,
        -0.076731354,
        0.034850594,
        0.06816987,
        -0.030740367,
        0.00038505212,
        0.05659121,
        -0.039674535,
        0.14872752,
        -0.028473355,
        -0.00791829,
        -0.5675016,
        0.02218855,
        -0.034207117,
        0.0019829122,
        -0.15359147,
        0.0049831034,
        -0.04684391,
        0.038168512,
        0.031071972,
        -0.009784036,
        0.025498277,
        -0.027783947,
        -0.021013137,
        -0.031613227,
        -0.035152316,
        -0.010126474,
        0.04399289
      ]
    },
    {
      "id": "worldsfair-speaker-105",
      "conference": "worldsfair",
      "name": "Sumanyu Sharma",
      "role": "Founder & CEO",
      "company": "Hamming AI",
      "twitter": "https://x.com/sumanyu",
      "talks": [
        {
          "title": "I Monitored Crime Audio. Voice Agents Scare Me More.",
          "description": "This talk reframes bad voice-agent calls as incident scenes and introduces a voice-agent forensics loop spanning transcript, waveform, latency waterfall, interruption points, ASR uncertainty, tool traces, system-of-record state, and outcomes. It focuses on monitoring, regression, and release-discipline for production voice systems.",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.013747779,
        0.052650098,
        0.019358953,
        0.018775959,
        0.024883203,
        -0.01708696,
        -0.04929411,
        -0.023449773,
        0.07294488,
        -0.23482965,
        -0.028821079,
        -0.055556774,
        -0.0091747865,
        -0.050576393,
        -0.099875964,
        -0.008403939,
        0.060952574,
        -0.10118929,
        -0.020255242,
        -0.051952276,
        -0.012760182,
        -0.05403008,
        -0.019233825,
        0.013325599,
        -0.0110172555,
        0.06917333,
        0.012029401,
        0.099194035,
        -0.07536255,
        0.45277682,
        -0.10831985,
        0.037130404,
        0.012651532,
        -0.05300164,
        0.06628443,
        -0.07057741,
        -0.0072345645,
        -0.019637367,
        0.004521189,
        0.06642889,
        -0.020603076,
        0.07576672,
        -0.035921767,
        0.03310001,
        0.05022449,
        -0.022177815,
        -0.036671814,
        -0.0023330548,
        -0.012739791,
        -0.011437138,
        -0.008511874,
        -0.050430745,
        -0.005484741,
        -0.008494876,
        0.067228034,
        0.038524583,
        0.029824965,
        -0.0054626707,
        0.10249452,
        -0.024532173,
        -0.12446203,
        0.013142223,
        0.026865484,
        0.037926912,
        0.033874214,
        -0.12449557,
        0.05067771,
        -0.061670836,
        -0.06463706,
        -0.014797057,
        0.025446774,
        -0.0006553926,
        -0.09478555,
        -0.0051385616,
        -0.050052278,
        -0.17077546,
        -0.0042504114,
        0.06540455,
        0.009670725,
        -0.00320946,
        -0.03204283,
        0.057796337,
        -0.0907118,
        -0.04589925,
        0.02027244,
        -0.05038662,
        -0.05358157,
        0.0115850475,
        0.02167392,
        0.028220072,
        0.032789424,
        -0.05154889,
        -0.05586809,
        0.0029559024,
        -0.04530702,
        0.00862293,
        -0.07097745,
        0.0036228134,
        -0.05671298,
        -0.004332434,
        0.00179716,
        7.493381e-7,
        0.02300122,
        0.037704714,
        -0.067291714,
        -0.027640361,
        0.059068095,
        -0.042349707,
        0.040342078,
        0.00055865647,
        -0.07931864,
        -0.6227856,
        0.04025752,
        -0.006518992,
        -0.072884314,
        -0.1534459,
        0.026144894,
        0.06780467,
        0.07604516,
        0.080337636,
        -0.014098325,
        0.051683042,
        0.0137508735,
        -0.036734898,
        -0.012436683,
        0.073148265,
        0.090194784,
        0.015237159
      ]
    },
    {
      "id": "worldsfair-speaker-106",
      "conference": "worldsfair",
      "name": "Ross Taylor",
      "role": "Co-founder & CEO",
      "company": "General Reasoning",
      "twitter": "https://x.com/rosstaylor90",
      "talks": [
        {
          "title": "General Reasoning for Long-Horizon Agent Models",
          "description": "Long-horizon agent models, reasoning loops, and the data/eval stack needed to make them reliable.",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0059902454,
        0.12437717,
        0.036484364,
        0.018989453,
        0.04871885,
        -0.07970081,
        -0.010868257,
        -0.0031887137,
        -0.01926603,
        -0.26212567,
        0.039136074,
        -0.09072336,
        -0.03366297,
        0.024039946,
        -0.051716454,
        0.026750315,
        0.067593224,
        -0.050528307,
        -0.04068779,
        -0.07498505,
        -0.06180445,
        -0.09124789,
        -0.015791133,
        0.035434287,
        0.007683427,
        0.116523504,
        -0.025412166,
        0.06568591,
        -0.10830601,
        0.42362157,
        -0.061097316,
        -0.013771105,
        0.04022536,
        -0.0076153437,
        0.0912273,
        -0.038766757,
        -0.048107978,
        -0.0110792145,
        0.03429596,
        0.026523493,
        -0.008988448,
        -0.04303867,
        -0.04741401,
        0.051896106,
        -0.012249099,
        0.0024431336,
        -0.0057992446,
        -0.011502721,
        -0.033566788,
        -0.025626956,
        -0.04134737,
        0.0065022185,
        0.05889817,
        -0.015850022,
        0.09226348,
        0.073677324,
        -0.008163718,
        -0.07190342,
        0.09147103,
        -0.05637163,
        -0.077803776,
        0.010731605,
        -0.00863558,
        -0.024748735,
        0.06766619,
        -0.06237939,
        0.021576492,
        -0.08658514,
        0.022932123,
        -0.07779045,
        0.09598829,
        0.041986465,
        -0.087694794,
        0.032032453,
        -0.07690696,
        -0.11427001,
        -0.044549216,
        0.051035758,
        -0.02094308,
        -0.01079253,
        0.06925095,
        0.04332685,
        -0.086337455,
        -0.036547843,
        0.029727403,
        -0.012454395,
        0.0006393594,
        0.0052320966,
        0.005147655,
        0.026949795,
        0.07574413,
        -0.040229414,
        -0.06334727,
        -0.01653859,
        -0.055480357,
        0.008336229,
        -0.0853227,
        0.04597252,
        -0.033681493,
        0.018742878,
        -0.04297908,
        0.03600747,
        0.046295222,
        -0.039590854,
        -0.059492547,
        0.007955349,
        0.017997563,
        -0.0392929,
        0.06430952,
        -0.08976771,
        -0.109319195,
        -0.61640406,
        -0.038384058,
        -0.043942217,
        -0.05310235,
        -0.15650713,
        0.033033423,
        0.030163703,
        0.01982311,
        0.029434577,
        -0.03733513,
        0.0821889,
        -0.006952593,
        0.025732232,
        0.010792841,
        0.015504493,
        0.1046718,
        0.007375849
      ]
    },
    {
      "id": "worldsfair-speaker-107",
      "conference": "worldsfair",
      "name": "Kanish Manuja",
      "role": "",
      "company": "Twilio",
      "twitter": "",
      "talks": [
        {
          "title": "Productionizing LLM Gateways: Architecture, Tradeoffs, and Hard Lessons from the Trenches",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.02233793,
        0.07067398,
        0.03176397,
        0.048079092,
        0.08369185,
        0.04166303,
        -0.09843387,
        0.026289126,
        0.04123094,
        -0.22156286,
        -0.06123812,
        -0.063293405,
        -0.0036584756,
        -0.041260242,
        -0.07931472,
        0.03111812,
        0.029399157,
        -0.06925984,
        -0.030465664,
        -0.09406746,
        -0.0377378,
        -0.014589923,
        0.018234847,
        -0.024736514,
        -0.042397413,
        0.04287156,
        -0.051917367,
        0.0013038564,
        -0.16049993,
        0.49156478,
        -0.079914376,
        -0.06892954,
        -0.0000071664713,
        -0.005902979,
        0.013132697,
        -0.009477359,
        0.015177148,
        0.006854489,
        0.03967581,
        0.006559292,
        -0.02550226,
        -0.009652239,
        -0.06191113,
        0.054948058,
        -0.06251865,
        0.0025290456,
        -0.03592144,
        0.0069595175,
        0.0020083813,
        -0.029721202,
        -0.07890157,
        0.022930646,
        0.012390514,
        -0.062171116,
        0.042700093,
        0.0447085,
        -0.03674754,
        -0.023141472,
        0.058985397,
        -0.00803893,
        -0.04845057,
        -0.004878898,
        0.002995334,
        -0.0354948,
        -0.026747981,
        -0.05399645,
        0.049309686,
        -0.09112703,
        -0.050318714,
        0.08737514,
        0.07676828,
        -0.044141438,
        -0.070222974,
        0.02934528,
        -0.040005106,
        -0.055137735,
        -0.0011923229,
        0.12872739,
        0.022240074,
        0.056235563,
        0.0059972154,
        0.0047836844,
        -0.09127513,
        -0.08510387,
        -0.019401725,
        -0.04487747,
        -0.04642051,
        0.05453487,
        -0.02305903,
        0.058736213,
        0.022307739,
        0.0300667,
        -0.039505113,
        -0.020159433,
        -0.0055403057,
        0.022986049,
        -0.027780266,
        0.04493085,
        -0.028280979,
        0.0075362558,
        0.003299757,
        -0.018523589,
        0.033739783,
        0.041138932,
        -0.07689,
        -0.015530965,
        0.02735668,
        -0.024205968,
        0.009328232,
        -0.035650723,
        -0.022919113,
        -0.64543825,
        -0.021106996,
        -0.0110102445,
        -0.05768349,
        -0.1367547,
        0.0139368465,
        -0.014126929,
        -0.014989188,
        0.039560445,
        0.027679216,
        0.061061185,
        -0.019707408,
        -0.04466556,
        0.013176696,
        0.03736796,
        0.01943292,
        -0.039630394
      ]
    },
    {
      "id": "worldsfair-speaker-108",
      "conference": "worldsfair",
      "name": "Kieran Klaassen",
      "role": "",
      "company": "Every/Cora",
      "twitter": "",
      "talks": [
        {
          "title": "The Era of Compound Engineering",
          "description": "Most codebases get harder to work with every year. Yours doesn't have to. **Compound Engineering** is a philosophy where each unit of work – every bug fix, every feature, every code review – makes the next one easier. This talk is about how that shift changes everything: from how fast you ship to how many engineers you actually need. --- At Every, we run five products with single-person engineering teams. That's not a headcount accident – it's a system. When I built [Cora](https://cora.computer), I wanted to find out how much one engineer could do with the right AI workflows. The answer became the **Compound Engineering** philosophy, now with 17k stars on GitHub. Traditional codebases accumulate complexity. Compound codebases accumulate capability. Bug fixes eliminate entire *categories* of future bugs. Patterns become tools. Over time, the codebase gets easier to understand, easier to modify, and easier to trust. **You'll walk away with:** - The mental model behind compound engineering - Concrete patterns for making every PR compound - How to scale output without scaling headcount",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.051850628,
        0.035594992,
        0.0016230966,
        -0.0038452423,
        -0.06117214,
        -0.04856907,
        -0.07144451,
        0.009974475,
        0.05052901,
        -0.20023403,
        -0.07446417,
        -0.035320207,
        0.054115236,
        0.0065881424,
        -0.0125741465,
        -0.012671476,
        0.10290105,
        -0.09400743,
        -0.094629936,
        -0.09671248,
        -0.013765972,
        -0.10775323,
        -0.041619685,
        0.08657797,
        -0.104444444,
        0.10304918,
        -0.02071173,
        -0.030834893,
        -0.09284508,
        0.5259668,
        -0.09984422,
        0.05108781,
        0.048804548,
        -0.029084051,
        0.026395863,
        -0.05611278,
        -0.06188928,
        0.053176712,
        0.060524203,
        0.074387275,
        -0.03256743,
        0.036557928,
        -0.05169649,
        0.030749226,
        -0.08766069,
        -0.017425645,
        -0.0438314,
        -0.08556308,
        0.025130989,
        -0.0684409,
        -0.09965531,
        0.02517401,
        0.00833768,
        -0.03711363,
        0.04735926,
        0.035168413,
        0.022021484,
        -0.026338007,
        0.056102533,
        -0.010567011,
        -0.052189074,
        -0.043381426,
        0.054326557,
        -0.0033157468,
        -0.04825002,
        -0.048298277,
        0.06371317,
        0.061222233,
        0.04342147,
        0.026461078,
        -0.008594367,
        0.030187014,
        -0.01833518,
        0.02930791,
        -0.042707454,
        -0.07078789,
        -0.064866826,
        0.0674383,
        -0.0011972651,
        -0.031288408,
        -0.0073504355,
        0.016044846,
        -0.043840364,
        -0.097083405,
        0.022044227,
        -0.032918043,
        -0.056625016,
        -0.012689303,
        0.010317024,
        0.058159493,
        0.011241761,
        -0.06348456,
        0.040012345,
        -0.007908016,
        -0.06320901,
        0.0029494418,
        0.035290547,
        -0.011347291,
        -0.054717693,
        0.007872044,
        -0.014491896,
        -0.0680906,
        0.042459853,
        -0.018183922,
        -0.054211557,
        0.0027941354,
        -0.009867646,
        -0.03911435,
        0.06515709,
        0.0026320003,
        0.003447766,
        -0.59929264,
        -0.018354641,
        0.01638459,
        0.03267488,
        -0.13896865,
        -0.037248056,
        -0.028211398,
        -0.0053167613,
        0.06419386,
        0.018084234,
        0.014975587,
        0.019725565,
        0.011529185,
        0.00043014289,
        0.016379228,
        0.039928168,
        0.011322971
      ]
    },
    {
      "id": "worldsfair-speaker-109",
      "conference": "worldsfair",
      "name": "Microsoft",
      "role": "",
      "company": "Microsoft",
      "twitter": "",
      "talks": [
        {
          "title": "Microsoft Presenting Expo Session 1",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:25pm-2:45pm",
          "type": "session"
        },
        {
          "title": "Microsoft Presenting Expo Session 2",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        },
        {
          "title": "Microsoft Presenting Expo Session 3",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03463261,
        0.05612632,
        -0.0106248325,
        -0.005715539,
        0.05292482,
        0.01995296,
        -0.07099251,
        -0.03218217,
        0.06360074,
        -0.27165648,
        -0.009327754,
        0.019904163,
        -0.011763423,
        -0.01417567,
        -0.021390634,
        0.030125419,
        0.038037866,
        -0.03320146,
        -0.101535305,
        -0.03984118,
        -0.04739347,
        -0.08337749,
        -0.05909559,
        0.012564736,
        -0.024283078,
        0.12264592,
        0.017350644,
        -0.025108099,
        -0.04168973,
        0.46938357,
        -0.009502371,
        0.012670288,
        -0.062446155,
        -0.024813585,
        0.026776858,
        -0.031209957,
        -0.062586844,
        -0.03523942,
        0.0426266,
        0.010922766,
        -0.03623368,
        0.08154673,
        -0.11289919,
        0.0075441394,
        0.021323882,
        -0.06223952,
        -0.10548875,
        -0.05152974,
        -0.004866386,
        -0.012293087,
        -0.033696894,
        -0.095945805,
        0.052904107,
        -0.04940044,
        0.043810215,
        0.027817681,
        -0.03381729,
        0.01594261,
        0.086388424,
        -0.035457283,
        -0.034820385,
        -0.04246387,
        0.0030891101,
        -0.013856231,
        -0.072178096,
        0.029264748,
        0.05671526,
        -0.04331487,
        -0.034052167,
        -0.022062533,
        0.07163793,
        0.010387661,
        -0.024124617,
        0.067623004,
        -0.018212289,
        -0.04620456,
        -0.067899846,
        0.03460853,
        0.013470052,
        0.010196387,
        0.053699937,
        0.09968232,
        -0.060908582,
        -0.037473757,
        -0.026375236,
        -0.0063019544,
        -0.05248029,
        -0.025249545,
        -0.0219545,
        0.05696449,
        -0.044986367,
        -0.027575448,
        -0.11564126,
        0.043174658,
        -0.03302805,
        0.0063719316,
        0.011690271,
        0.06727003,
        -0.016981378,
        -0.019072367,
        0.0065082503,
        -0.005833721,
        -0.030394165,
        -0.0056054234,
        -0.025473665,
        -0.09081192,
        -0.018767787,
        0.017937114,
        0.15400667,
        -0.009658676,
        -0.07483047,
        -0.6102416,
        -0.0326101,
        -0.026601084,
        -0.023465017,
        -0.14872003,
        0.04074271,
        -0.09751719,
        0.09915408,
        0.07864992,
        -0.013589009,
        0.049280435,
        -0.04106828,
        0.021222005,
        -0.0034478346,
        0.033578135,
        0.039857905,
        -0.050438598
      ]
    },
    {
      "id": "worldsfair-speaker-110",
      "conference": "worldsfair",
      "name": "Eric Zhu",
      "role": "",
      "company": "Alibaba",
      "twitter": "",
      "talks": [
        {
          "title": "QwenPaw: building AI that you can trust",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.07023077,
        0.036518738,
        -0.05110404,
        0.013094886,
        0.010804,
        -0.08043389,
        -0.05881582,
        -0.007291769,
        0.028275024,
        -0.24190496,
        0.023530873,
        -0.04802101,
        0.022178883,
        0.0657217,
        -0.010837894,
        0.027617471,
        0.039160006,
        -0.06992244,
        -0.052216005,
        -0.024571529,
        -0.034450397,
        -0.07240937,
        -0.04007225,
        0.0012543756,
        -0.055080842,
        0.12956327,
        0.034851525,
        0.060087148,
        -0.078890786,
        0.44355264,
        -0.08930052,
        0.0057699275,
        0.025297482,
        0.008531802,
        0.04462616,
        -0.057628147,
        -0.011620169,
        -0.064326614,
        0.03940993,
        0.080815315,
        -0.01964849,
        -0.00438109,
        -0.062267143,
        0.038963743,
        -0.014484315,
        -0.015396125,
        -0.026685998,
        -0.046488572,
        0.014976049,
        0.049673706,
        -0.009758457,
        -0.0067721005,
        0.02596429,
        -0.042983323,
        0.0034773885,
        0.034960184,
        -0.0140754795,
        -0.00025481152,
        0.06112657,
        0.0012665283,
        -0.12339415,
        -0.031337075,
        0.013172549,
        0.0023775701,
        -0.015393183,
        -0.026332412,
        0.03806175,
        -0.06987213,
        -0.030724842,
        -0.022263186,
        -0.032969527,
        -0.035356015,
        -0.033929866,
        0.016921828,
        -0.09592002,
        -0.052555222,
        0.023916993,
        0.114731625,
        -0.0663426,
        -0.068368755,
        0.07356258,
        0.023724556,
        -0.059631523,
        -0.116318665,
        0.012208558,
        0.03827066,
        -0.08193127,
        0.03074388,
        -0.021237079,
        0.0824446,
        -0.04925683,
        0.015947375,
        -0.04062726,
        -0.006602441,
        -0.028082231,
        0.051757574,
        -0.06698861,
        0.078661725,
        0.006826441,
        0.046162132,
        0.0036909212,
        0.013947631,
        0.044211034,
        -0.029229788,
        -0.040264983,
        -0.03625529,
        0.031034436,
        -0.002330319,
        0.028192598,
        -0.016388612,
        -0.05598131,
        -0.6349612,
        -0.032925952,
        -0.009159941,
        -0.09399975,
        -0.15360445,
        -0.048924897,
        0.00057098386,
        0.07321989,
        0.12958774,
        -0.06373725,
        0.11351454,
        0.037145384,
        0.023118634,
        0.039117638,
        0.06516139,
        -0.010789106,
        0.021011688
      ]
    },
    {
      "id": "worldsfair-speaker-111",
      "conference": "worldsfair",
      "name": "Andrew Dai",
      "role": "",
      "company": "Elorian AI",
      "twitter": "",
      "talks": [
        {
          "title": "The Best Models Still Reason Like Toddlers",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.0137323635,
        0.08719577,
        -0.014250035,
        0.00051245856,
        0.010748899,
        0.028989471,
        -0.068932846,
        0.0035573666,
        0.043727938,
        -0.25772664,
        -0.029222192,
        -0.10642928,
        -0.0026453396,
        0.046503607,
        -0.06372372,
        0.03840418,
        0.039716452,
        -0.08302259,
        -0.09508121,
        -0.037500817,
        -0.010126503,
        -0.03360552,
        -0.02933458,
        0.067581415,
        -0.031016542,
        0.120352544,
        0.013307576,
        0.04812783,
        -0.09603605,
        0.43266955,
        -0.08496267,
        -0.008219685,
        0.014098183,
        -0.022952493,
        0.055633128,
        -0.11552016,
        -0.03597028,
        -0.030170558,
        0.07723708,
        0.07062846,
        -0.016396783,
        -0.03561285,
        -0.086765505,
        0.07951494,
        -0.01998971,
        0.06791343,
        -0.08341767,
        -0.055574138,
        -0.034363214,
        0.048857823,
        -0.064468265,
        0.018288469,
        0.07970151,
        -0.06251376,
        0.010851064,
        0.030347342,
        -0.05373093,
        -0.016898438,
        0.07802821,
        -0.029401867,
        -0.1301894,
        -0.01054549,
        -0.02159939,
        0.00335528,
        0.013012998,
        -0.06008259,
        0.027025022,
        -0.07354915,
        0.041495375,
        -0.015705742,
        0.04895344,
        0.05041313,
        -0.019908141,
        0.023722377,
        -0.12813203,
        -0.03355742,
        -0.033432536,
        0.09042591,
        -0.06497317,
        0.01832706,
        0.07249081,
        0.022319667,
        -0.03713907,
        -0.0030357204,
        -0.017183956,
        -0.07210054,
        -0.025772689,
        0.030818155,
        -0.018854722,
        0.12382731,
        -0.030779094,
        -0.092162184,
        -0.065498635,
        -0.051487297,
        -0.0471501,
        0.02356398,
        -0.07817347,
        0.03665195,
        0.012413816,
        -0.002451423,
        0.03660358,
        0.0197166,
        0.09852082,
        -0.033858944,
        -0.0674994,
        -0.09484076,
        -0.0048933863,
        -0.037493456,
        0.050564114,
        0.012248463,
        0.0045517986,
        -0.60703814,
        -0.010775214,
        -0.05840461,
        -0.009180215,
        -0.16250524,
        -0.005672244,
        -0.010543738,
        0.03849328,
        0.012328669,
        -0.014287901,
        0.023261188,
        0.030558318,
        -0.0015873106,
        0.025362723,
        0.031671826,
        0.0534733,
        0.019500239
      ]
    },
    {
      "id": "worldsfair-speaker-112",
      "conference": "worldsfair",
      "name": "Yuval Belfer",
      "role": "Senior Developer Advocate",
      "company": "AI21 Labs",
      "twitter": "",
      "talks": [
        {
          "title": "Stop Chunking Like It's 2022",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        },
        {
          "title": "Two Bugs That Hid in Plain Sight: A vLLM Debugging Detective Story",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.047264177,
        -0.004325509,
        0.0031754961,
        0.0078061447,
        -0.0038682518,
        -0.032925308,
        -0.08004987,
        0.02591964,
        0.05444344,
        -0.2178991,
        -0.033611152,
        -0.08516936,
        -0.009886756,
        0.01298774,
        -0.033281963,
        0.044157036,
        0.1322813,
        -0.09075083,
        -0.102017686,
        -0.14127402,
        -0.03514744,
        -0.11521282,
        -0.025255486,
        0.005512198,
        -0.032290168,
        0.07828273,
        0.0016046519,
        0.048438102,
        -0.09982031,
        0.45579222,
        -0.15584368,
        -0.0063848346,
        0.014085535,
        -0.087492,
        0.023943862,
        -0.09896237,
        -0.028788447,
        -0.012646768,
        -0.043502335,
        0.0686793,
        -0.007113486,
        0.0022032717,
        -0.014913756,
        0.0041314014,
        -0.057956994,
        -0.04578903,
        -0.10386042,
        0.0040077595,
        -0.032358207,
        -0.05124587,
        -0.027085248,
        -0.0073777037,
        0.08335598,
        -0.017847437,
        -0.0126827825,
        0.017666992,
        0.031784054,
        -0.08222997,
        0.10156399,
        -0.010436141,
        -0.08514884,
        0.0010490778,
        -0.05775932,
        0.0013007559,
        0.028553387,
        -0.04493709,
        0.033404365,
        -0.104347944,
        -0.06299728,
        -0.008853841,
        -0.009530817,
        0.017237168,
        -0.04170254,
        0.013082063,
        -0.059531,
        -0.06839834,
        -0.055240195,
        0.10453028,
        -0.032960493,
        -0.0025426575,
        0.03999036,
        0.08848352,
        -0.08323332,
        0.014287373,
        0.03349452,
        0.006369485,
        -0.07792735,
        0.085411735,
        -0.0032881927,
        0.091125086,
        0.057000984,
        -0.07470558,
        -0.0726219,
        0.07602785,
        -0.071348414,
        0.062391613,
        -0.025417455,
        0.077028625,
        0.025927303,
        -0.025242358,
        0.022236953,
        -0.010882349,
        -0.046583556,
        -0.020391218,
        -0.005890769,
        -0.060366705,
        -0.019013708,
        0.016359614,
        0.07865902,
        0.02188116,
        -0.07044246,
        -0.56952614,
        0.03679222,
        -0.08581638,
        -0.058994576,
        -0.15017338,
        -0.028296705,
        -0.055994775,
        -0.0076408973,
        0.07909095,
        0.0057564024,
        0.03500719,
        0.0036997704,
        0.037838377,
        0.015483963,
        0.07269101,
        0.05729417,
        -0.022512287
      ]
    },
    {
      "id": "worldsfair-speaker-113",
      "conference": "worldsfair",
      "name": "Bo Li",
      "role": "",
      "company": "EliseAi",
      "twitter": "",
      "talks": [
        {
          "title": "Realtime Voice Agents with Frontier Intelligence",
          "description": "A deep dive into an EliseAI voice-agent harness that orchestrates multiple models to achieve realtime latency without sacrificing intelligence. The talk covers speculative transcription, async background tool injection, and TTS prefix caching/infilling to reduce latency while preserving capability.",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.070616715,
        0.08590027,
        0.032205936,
        0.060960323,
        0.0055226297,
        0.032959756,
        -0.09932697,
        0.023316216,
        0.009620862,
        -0.22789766,
        -0.011341276,
        -0.06018756,
        -0.04640016,
        0.022465855,
        -0.08102497,
        -0.0019559595,
        0.052102223,
        -0.044119574,
        -0.09089249,
        -0.045871593,
        -0.042601466,
        -0.07992819,
        -0.0145253,
        0.004296969,
        -0.034091387,
        0.047534503,
        -0.03838776,
        0.092322804,
        -0.088390484,
        0.41314468,
        -0.06685996,
        0.054633,
        0.06715781,
        -0.023924742,
        0.082557425,
        -0.0655371,
        -0.04275779,
        -0.036141526,
        0.066496395,
        0.07324802,
        0.0032152813,
        0.028830612,
        -0.027843952,
        0.036126975,
        -0.056289207,
        -0.009230793,
        -0.04579323,
        -0.050600212,
        -0.044722583,
        0.032634083,
        -0.059015214,
        0.006883593,
        -0.00591877,
        -0.010243006,
        0.03325199,
        0.043614585,
        0.0008442198,
        -0.08429096,
        0.1149037,
        0.05042697,
        -0.12226789,
        0.013478388,
        0.07200313,
        -0.004761184,
        0.0559111,
        -0.062980816,
        -0.019924989,
        -0.108353846,
        -0.02800866,
        -0.036763217,
        0.02982255,
        0.042073604,
        -0.04357072,
        -0.0095133865,
        -0.071763545,
        -0.1432463,
        -0.034068994,
        0.128268,
        -0.01752014,
        0.02907,
        -0.009364492,
        0.13883008,
        -0.10100162,
        -0.053592063,
        0.031194467,
        -0.020922486,
        -0.11018607,
        -0.009789352,
        -0.029180514,
        0.059905097,
        0.065273315,
        -0.010036428,
        -0.027740141,
        0.0028811174,
        -0.06102444,
        0.029061588,
        -0.043869358,
        0.050669912,
        -0.012634747,
        0.06553348,
        0.028528666,
        0.027502706,
        0.060634386,
        -0.018854257,
        -0.035129428,
        -0.0502025,
        0.037363295,
        -0.004888559,
        0.08271394,
        0.012241925,
        -0.06347885,
        -0.62003195,
        -0.0049605016,
        -0.07625754,
        -0.022798413,
        -0.15625419,
        0.036573444,
        0.02413462,
        0.04435346,
        0.04159301,
        -0.036032625,
        0.06635179,
        -0.0040330486,
        -0.044122715,
        -0.0034683254,
        0.01870876,
        0.01842302,
        0.003258784
      ]
    },
    {
      "id": "worldsfair-speaker-114",
      "conference": "worldsfair",
      "name": "Nick Heiner",
      "role": "",
      "company": "Surge",
      "twitter": "",
      "talks": [
        {
          "title": "Beyond the Benchmark: the New Frontier of Enterprise AI Reliability",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.016031824,
        0.047670443,
        0.06924462,
        0.08974522,
        0.03841281,
        -0.019920338,
        -0.1127401,
        -0.038963173,
        0.025468083,
        -0.24703285,
        -0.060133,
        -0.022892267,
        -0.03547231,
        0.056474738,
        -0.08299698,
        0.002650689,
        0.16267592,
        -0.09345378,
        -0.051243395,
        -0.09335463,
        0.037182655,
        -0.11031119,
        0.056506023,
        -0.008766421,
        -0.07886873,
        0.061952017,
        -0.017268844,
        0.059913274,
        -0.10043328,
        0.45958477,
        -0.08071887,
        -0.020787112,
        0.03638457,
        -0.01749407,
        0.030690042,
        -0.018609507,
        -0.031499635,
        -0.02246537,
        -0.04712643,
        0.13339883,
        -0.0059377463,
        0.03445572,
        -0.042710148,
        -0.00044306548,
        -0.04112566,
        0.009694848,
        -0.07328753,
        -0.010053466,
        -0.010630824,
        -0.023600815,
        -0.08618312,
        0.060040828,
        0.018944036,
        0.036408253,
        0.041815005,
        0.029346533,
        -0.021594778,
        -0.02150059,
        0.12357293,
        -0.056032844,
        -0.12050279,
        -0.011648616,
        0.0010785956,
        0.033011634,
        0.02373743,
        -0.10188735,
        0.082591206,
        -0.036641918,
        -0.03432641,
        -0.027569266,
        -0.020613058,
        -0.057562225,
        -0.05573408,
        -0.043827306,
        -0.0983297,
        -0.046520058,
        0.0063312016,
        0.06390151,
        -0.020943083,
        -0.05323993,
        0.027720058,
        0.010066709,
        -0.045601264,
        -0.06343145,
        -0.04262889,
        -0.02519857,
        -0.029964568,
        -0.014823677,
        -0.0034945535,
        0.0768312,
        0.0935306,
        -0.053987343,
        -0.017653825,
        0.018480357,
        -0.046774883,
        -0.041256905,
        -0.09344624,
        -0.022423046,
        -0.0032984822,
        0.016750319,
        -0.024863219,
        -0.0601728,
        0.040531933,
        -0.021948487,
        -0.07110697,
        0.003677885,
        -0.02990909,
        -0.0693946,
        -0.0004915247,
        0.0021183947,
        -0.059439737,
        -0.58462435,
        0.016318766,
        -0.058093823,
        -0.00043331695,
        -0.14401974,
        0.026281925,
        0.055885136,
        0.04516785,
        0.029112998,
        0.027347561,
        0.06236693,
        -0.010216442,
        0.032972004,
        -0.0023350616,
        0.07271884,
        0.040309373,
        0.04957007
      ]
    },
    {
      "id": "worldsfair-speaker-115",
      "conference": "worldsfair",
      "name": "Clare Liguori",
      "role": "Senior Principal SWE",
      "company": "Amazon Web Services",
      "twitter": "https://x.com/clare_liguori",
      "talks": [
        {
          "title": "From AI-Assisted to AI-Native: Building a Frontier Development Team",
          "description": "When features that took two weeks now ship in an afternoon, the bottleneck shifts from writing code to making decisions. Frontier teams have discovered this firsthand, achieving 3-10x productivity gains by fundamentally rethinking how developers work with AI agents. This talk covers the practices that separate frontier teams from those who merely \"sprinkle\" AI on their existing workflows: running agents asynchronously for hours, investing in comprehensive agent steering files, enabling local integration testing for agent self-correction, and automating everything from coding to operations to documentation. You'll learn how teams at Amazon slowed down to speed up, the temporary productivity dips they accepted, and the organizational changes required to sustain this velocity.",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        },
        {
          "title": "The Infinite Context Window Is a Myth: Context Engineering for AI Agents",
          "description": "Large context windows have become a popular answer to the growing complexity of AI agents. When agents lose track of details, forget prior decisions, or degrade in reasoning quality, the instinct is often to add more tokens. In practice, this rarely fixes the problem and often makes it worse. Bigger context windows increase cost and latency, introduce noise, and amplify failure modes like lost-in-the-middle effects, context collapse, and brittle summarization. This talk argues that the real challenge is not context size, but context engineering. In this session, we will explore practical context engineering techniques for building AI agents that reason reliably over time without relying on ever-larger context windows. Starting from a stateless agent, we will walk through progressively more advanced strategies, including short-term and long-term memory, conversation curation policies, retrieval-augmented generation, and tool-driven context injection. We will examine common failure modes such as context pollution from tool outputs, brevity bias during summarization, and reasoning degradation as conversations grow, and show concrete ways to mitigate them. The talk is grounded in real agent implementations using the Strands Agents SDK and Amazon Bedrock AgentCore, but the principles apply broadly to any agent framework. This session is intended for engineers building AI agents beyond simple chatbots who want practical techniques they can apply immediately.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0338562,
        0.031576443,
        0.057879236,
        0.025474317,
        0.06114482,
        -0.028360497,
        -0.08847733,
        -0.04633235,
        0.025492875,
        -0.1950112,
        -0.026593678,
        -0.093143374,
        0.048077267,
        0.0044354275,
        -0.06919717,
        -0.0038155396,
        0.072831854,
        -0.05059353,
        -0.07348441,
        -0.06397907,
        -0.04508476,
        -0.06582491,
        -0.016656674,
        0.09423428,
        -0.07233496,
        0.083598696,
        0.039293982,
        0.027758136,
        -0.12199292,
        0.45166603,
        -0.12415217,
        -0.0061854525,
        0.020214878,
        -0.033554055,
        0.09835276,
        -0.076682605,
        0.002210841,
        -0.028961055,
        0.04829261,
        0.021303626,
        0.05385334,
        0.029540092,
        -0.05279947,
        0.05466534,
        -0.052553706,
        0.011228888,
        -0.06893869,
        -0.027197216,
        -0.030102411,
        0.03870763,
        -0.07090377,
        0.017415216,
        0.05600327,
        0.061781876,
        0.05267624,
        0.043389235,
        0.014429947,
        -0.055746414,
        0.13262576,
        -0.016067054,
        -0.085176475,
        -0.035999924,
        -0.022839144,
        0.017620636,
        0.026288062,
        -0.07605452,
        0.04089455,
        -0.06559582,
        -0.024108965,
        -0.035196293,
        0.006321696,
        0.012190007,
        -0.03156282,
        0.02454025,
        -0.030343274,
        -0.109091975,
        -0.035944086,
        0.05758228,
        -0.00671797,
        0.02457618,
        0.017857684,
        0.07893115,
        -0.08700222,
        -0.08340271,
        0.049010962,
        0.001683208,
        -0.05396177,
        0.033122554,
        -0.11737502,
        0.054837227,
        0.051839214,
        -0.044973683,
        -0.08386497,
        0.023713838,
        -0.08797127,
        -0.036121797,
        -0.04073718,
        0.02075613,
        0.032256912,
        0.06430978,
        0.00031063482,
        -0.04866943,
        0.032281782,
        -0.031871863,
        -0.008072918,
        0.011828198,
        -0.032766595,
        0.005567823,
        0.09221937,
        -0.0051093553,
        -0.04537611,
        -0.6157468,
        -0.0014499687,
        0.02058684,
        -0.05831685,
        -0.13807069,
        0.05206424,
        -0.06458135,
        0.0433073,
        0.056724004,
        -0.024667813,
        0.093973115,
        -0.034609545,
        -0.001754564,
        -0.043606542,
        -0.0038967023,
        0.06066722,
        -0.015964467
      ]
    },
    {
      "id": "worldsfair-speaker-116",
      "conference": "worldsfair",
      "name": "Niels Rogge",
      "role": "",
      "company": "Hugging Face",
      "twitter": "https://x.com/NielsRogge",
      "talks": [
        {
          "title": "How I automate my own job at Hugging Face using agents",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.033400156,
        0.008968525,
        0.047455575,
        0.025378551,
        0.019641237,
        0.0021018754,
        -0.07124575,
        0.042202543,
        0.024217414,
        -0.20290165,
        -0.0639241,
        -0.011924122,
        0.039516896,
        0.022781279,
        -0.07625015,
        0.02891221,
        0.12713099,
        -0.08024666,
        -0.06361203,
        -0.065465696,
        -0.05798268,
        -0.053957913,
        0.013466587,
        0.090691276,
        -0.004015179,
        0.08968,
        -0.050882537,
        0.023992276,
        -0.12549353,
        0.50985944,
        -0.11186488,
        0.015628008,
        0.020732803,
        -0.014595339,
        0.034951746,
        -0.06346271,
        -0.029211929,
        0.011786906,
        -0.025633067,
        0.08551556,
        -0.008887077,
        0.018234968,
        -0.035931915,
        -0.003981288,
        -0.051538747,
        0.06923312,
        -0.04410305,
        0.02824018,
        -0.018956987,
        -0.036872175,
        -0.11235815,
        0.042541146,
        0.045832753,
        -0.04578747,
        -0.016696114,
        0.017990699,
        0.03600684,
        -0.105357885,
        0.10625069,
        0.014513761,
        -0.05629045,
        -0.06376619,
        -0.034156196,
        -0.014576248,
        -0.003261089,
        -0.012532543,
        0.07136362,
        -0.10388273,
        0.019800724,
        -0.06454706,
        0.0058530844,
        -0.022498664,
        -0.020846331,
        0.017252132,
        -0.08412175,
        -0.050040014,
        0.017490761,
        0.05674711,
        -0.03790966,
        -0.040168446,
        -0.024197148,
        0.0009791155,
        -0.03802739,
        -0.033877324,
        0.07056951,
        -0.059179004,
        -0.07340792,
        -0.021233156,
        0.0029628482,
        0.05919048,
        0.09461502,
        -0.027918488,
        0.0027123585,
        0.01983412,
        -0.039729938,
        -0.007790783,
        -0.031245334,
        0.036452234,
        0.042644773,
        0.063553855,
        -0.053620014,
        -0.038812883,
        0.038799256,
        -0.08935371,
        -0.04890478,
        -0.017710524,
        -0.005874037,
        -0.004126773,
        0.03599306,
        -0.017732374,
        -0.016695945,
        -0.59819376,
        -0.05389093,
        -0.04402915,
        -0.058674335,
        -0.18089557,
        0.017584065,
        -0.0027365114,
        -0.02671869,
        0.06141332,
        0.022679228,
        0.053166233,
        0.010983763,
        0.014422856,
        0.0033751775,
        0.031110276,
        0.04706863,
        -0.020040633
      ]
    },
    {
      "id": "worldsfair-speaker-117",
      "conference": "worldsfair",
      "name": "Vaibhav Gupta",
      "role": "Founder and CEO",
      "company": "Boundary",
      "twitter": "",
      "talks": [
        {
          "title": "fighting slop with slop",
          "description": "We haven't done a code review in two years. The last time I read every line of code in a PR was about six months ago. And we build a programming language with a runtime meant to replace V8. This is real engineering: compiler internals, runtime behavior, type systems, codegen, concurrency semantics, and FFIs across multiple languages. The thing that makes this possible is a technique we call \"fight slop with slop\" - every line of code is analyzed in depth by a sprawling toolchain of custom visualizers, linters, test snapshots and a whole bunch more. While the core language VM code has super high standards, a lot of these meta-tools are mostly vibe-coded. I'll dive deep into all the tactical things we've built, and how to adopt \"fight slop with slop\" in your own team",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.078129336,
        0.06469335,
        0.10740921,
        0.019531822,
        -0.042652927,
        0.003175802,
        -0.14784664,
        -0.009079283,
        0.11441349,
        -0.23290803,
        -0.12030029,
        -0.05733317,
        0.03923619,
        -0.03254134,
        -0.05310203,
        0.05087723,
        0.0785115,
        -0.034201752,
        -0.023303706,
        -0.0809344,
        -0.024872221,
        -0.01780147,
        -0.025571,
        0.035562746,
        -0.030829575,
        0.042665232,
        0.024107805,
        0.0462146,
        -0.074198596,
        0.46552226,
        -0.05842202,
        0.0152406525,
        0.018377455,
        -0.05651387,
        0.020078398,
        -0.028612083,
        -0.041313536,
        0.062780015,
        0.0054649063,
        0.07384816,
        -0.061346136,
        0.03707583,
        -0.053270735,
        -0.015137256,
        0.027408386,
        -0.018371668,
        -0.066160835,
        0.0043441253,
        0.0061507453,
        -0.06812817,
        -0.07126231,
        -0.032573048,
        0.0755396,
        -0.012307043,
        -0.0060591716,
        0.059044268,
        -0.038860805,
        -0.02778044,
        0.02928185,
        -0.07529223,
        -0.068349846,
        -0.033002123,
        0.055527084,
        0.0056970674,
        0.0339517,
        -0.0404981,
        0.0747036,
        -0.04186022,
        -0.003139488,
        0.041953,
        0.02202217,
        0.03836785,
        -0.05657056,
        0.027921116,
        -0.031711034,
        -0.10443329,
        -0.048797045,
        0.035184417,
        -0.016142836,
        -0.024924083,
        0.06552208,
        0.008257662,
        -0.05762539,
        -0.012129704,
        0.03455745,
        -0.07156978,
        -0.018516298,
        -0.012811652,
        0.02303618,
        0.09817162,
        0.033694524,
        -0.094913386,
        -0.019453121,
        0.08785954,
        -0.03997274,
        0.08473646,
        -0.020474302,
        0.05285243,
        -0.061732873,
        0.042864073,
        0.04260555,
        0.0067295996,
        0.03716726,
        0.023197455,
        0.026922623,
        -0.009482048,
        -0.032268535,
        -0.05417502,
        -0.030620346,
        -0.006226203,
        -0.0739746,
        -0.6185922,
        -0.004927911,
        0.066842526,
        -0.028502457,
        -0.13748688,
        0.05423014,
        0.008888556,
        0.039059196,
        0.07829835,
        -0.02481387,
        0.073956154,
        -0.062225,
        -0.0005616106,
        -0.023512652,
        0.07230928,
        0.03725157,
        -0.014036221
      ]
    },
    {
      "id": "worldsfair-speaker-118",
      "conference": "worldsfair",
      "name": "Ben Guo",
      "role": "",
      "company": "Zo Computer",
      "twitter": "https://x.com/0thernet",
      "talks": [
        {
          "title": "Everyone Gets A Software Company",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.099091746,
        0.035097912,
        0.016626766,
        -0.016426621,
        -0.05283504,
        -0.03835207,
        -0.08591174,
        0.04699843,
        0.036124036,
        -0.2591904,
        -0.07005009,
        -0.06934993,
        0.01879495,
        0.07850089,
        -0.057296418,
        0.041451376,
        0.026876416,
        -0.073575236,
        -0.06646228,
        -0.010635093,
        -0.052832358,
        -0.09997521,
        -0.05526461,
        0.013178257,
        -0.045461103,
        0.10461072,
        0.03328786,
        -0.03996053,
        -0.053938046,
        0.4296804,
        -0.012860722,
        0.068459876,
        -0.021127122,
        0.010699753,
        0.058849253,
        -0.014773147,
        0.015579951,
        0.05788571,
        0.029762791,
        0.05468857,
        -0.0695899,
        0.081235394,
        -0.0801676,
        0.0104072355,
        0.014894241,
        0.06172559,
        -0.07499886,
        -0.08473813,
        0.024094358,
        -0.027570195,
        -0.05435534,
        0.02152579,
        0.03512047,
        -0.052399114,
        -0.00015927253,
        -0.0322627,
        0.009067848,
        -0.037023883,
        0.055550776,
        0.07849685,
        -0.17943719,
        -0.059697445,
        -0.001528197,
        0.022905033,
        -0.032629415,
        -0.0020574643,
        0.0061647114,
        0.0036211237,
        -0.050172765,
        -0.04037006,
        0.082291,
        0.024821552,
        -0.019633954,
        0.13025343,
        -0.10201923,
        -0.0034281558,
        -0.006734466,
        0.030256722,
        0.006522795,
        0.0050267717,
        0.05478495,
        -0.002177956,
        -0.06854553,
        -0.06292911,
        -0.028369,
        -0.014823187,
        -0.039183423,
        0.034730174,
        0.007750367,
        0.08869371,
        0.06812294,
        0.0007257791,
        -0.03597925,
        0.06295862,
        -0.061645664,
        -0.011174625,
        -0.05120985,
        0.07179813,
        -0.067548394,
        0.023601804,
        -0.058412585,
        -0.039295707,
        0.03398763,
        -0.00816828,
        -0.06390561,
        -0.09158322,
        -0.060443725,
        -0.056281857,
        0.045453515,
        0.014187309,
        0.020545272,
        -0.6093586,
        -0.0504792,
        -0.030188978,
        0.032058768,
        -0.13915953,
        0.041289404,
        -0.038203254,
        0.024600914,
        0.0862133,
        0.0021240849,
        0.024498133,
        0.07020805,
        0.09669119,
        -0.047441553,
        0.039086916,
        -0.043993797,
        -0.009628923
      ]
    },
    {
      "id": "worldsfair-speaker-119",
      "conference": "worldsfair",
      "name": "Rob Cheung",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Everyone Gets A Software Company",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.063688,
        0.048094817,
        -0.035495695,
        -0.00980973,
        -0.0015729949,
        -0.03427772,
        -0.06197617,
        0.065496825,
        0.030701393,
        -0.2472055,
        -0.10458088,
        -0.06594239,
        0.025232682,
        0.013575391,
        -0.03625869,
        0.061730668,
        0.05767825,
        -0.09952277,
        -0.1081695,
        -0.057251062,
        -0.050904557,
        -0.11933538,
        -0.048054464,
        0.070859686,
        -0.07467479,
        0.10604208,
        0.019373793,
        -0.007075254,
        -0.112771824,
        0.46266672,
        -0.063540645,
        0.03566457,
        -0.013132096,
        0.0112313,
        0.10743241,
        -0.0016318387,
        0.061616074,
        0.06256419,
        0.082200035,
        0.13205622,
        -0.029460615,
        0.006503854,
        -0.022494987,
        0.0026380597,
        0.045881987,
        0.05086784,
        -0.09312744,
        -0.08695189,
        0.022122422,
        -0.0036478955,
        -0.10975828,
        0.026112238,
        0.016736213,
        -0.009688147,
        0.040210385,
        0.007804879,
        -0.024485812,
        -0.008968438,
        0.042852852,
        -0.0045187892,
        -0.11141333,
        -0.052065574,
        -0.014203557,
        -0.021071985,
        -0.08466681,
        -0.009183997,
        0.029858628,
        -0.022775054,
        0.031185225,
        -0.05009679,
        0.07538812,
        -0.029619051,
        -0.009468834,
        0.043513753,
        -0.097966745,
        -0.020184109,
        -0.006829617,
        0.054913707,
        0.025945049,
        -0.03742333,
        0.052291613,
        0.020837191,
        -0.047012825,
        -0.031601343,
        -0.02439899,
        0.0037517971,
        -0.06430453,
        0.001705543,
        0.034713674,
        0.11487693,
        0.054034445,
        -0.023626545,
        0.0014990258,
        0.005850285,
        -0.035159256,
        -0.0057195025,
        -0.009111599,
        0.1210239,
        -0.08791214,
        0.038485352,
        -0.10493093,
        -0.002128399,
        -0.0016843805,
        -0.005957562,
        -0.026507417,
        -0.067418285,
        -0.040035665,
        -0.1103638,
        0.07625069,
        -0.024418011,
        0.015634432,
        -0.56695455,
        -0.03977864,
        -0.003340582,
        0.016005596,
        -0.11423137,
        0.03142859,
        -0.100626595,
        0.022200014,
        0.10561707,
        0.027675007,
        0.014548015,
        0.05513328,
        0.029389525,
        -0.06735699,
        0.0029730923,
        0.017194383,
        -0.05146775
      ]
    },
    {
      "id": "worldsfair-speaker-120",
      "conference": "worldsfair",
      "name": "Jia-Bin Huang",
      "role": "Associate Professor",
      "company": "University of Maryland",
      "twitter": "",
      "talks": [
        {
          "title": "Jia-Bin Huang",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.04264438,
        -0.020271683,
        -0.0065776566,
        -0.03169789,
        0.018993802,
        -0.12108074,
        -0.071802124,
        -0.050276704,
        0.08267475,
        -0.24346144,
        -0.016729198,
        -0.00009006797,
        0.018835047,
        0.034491293,
        -0.06109466,
        0.04873993,
        -0.009492098,
        -0.02445779,
        -0.07291624,
        -0.10234036,
        -0.0020430232,
        -0.068613976,
        -0.022457473,
        0.069615245,
        -0.030848417,
        0.11751678,
        0.040837288,
        0.032058105,
        -0.12158988,
        0.40352547,
        -0.09794897,
        0.009092254,
        -0.01253689,
        -0.01130033,
        0.06534067,
        -0.10747378,
        -0.10114922,
        0.020591123,
        0.038879458,
        0.07066582,
        -0.05653712,
        -0.0036030964,
        -0.080358446,
        0.055759013,
        0.00578537,
        -0.011370723,
        -0.021530276,
        0.005429094,
        0.015517601,
        0.06773539,
        -0.040225893,
        -0.039572384,
        0.046580017,
        -0.0363662,
        0.0023762395,
        -0.014596549,
        0.00442719,
        0.009372638,
        0.10291668,
        -0.017465927,
        -0.13398097,
        0.01605282,
        0.059855044,
        -0.005685902,
        0.014562174,
        -0.030869767,
        0.07288518,
        0.017467499,
        0.009323133,
        0.04655211,
        0.046792015,
        0.046615526,
        -0.025694946,
        0.020126233,
        -0.10051694,
        -0.003631651,
        -0.016666574,
        0.031139152,
        -0.040846355,
        -0.015368441,
        0.06427738,
        0.064653784,
        -0.03624492,
        -0.10255789,
        0.026940737,
        -0.021984372,
        -0.122023866,
        0.030739106,
        -0.04364749,
        0.1414771,
        0.011219342,
        -0.06711553,
        -0.015861671,
        0.042886082,
        -0.04001896,
        0.0694551,
        -0.058649894,
        0.07812778,
        0.068580255,
        0.08094461,
        0.0059443926,
        0.03436794,
        0.016969569,
        -0.02338009,
        -0.048932143,
        0.0143019045,
        0.026152832,
        -0.031107632,
        0.08806831,
        -0.03134738,
        -0.047042206,
        -0.6288942,
        -0.014138635,
        0.0041537313,
        -0.014925719,
        -0.15921314,
        -0.056926906,
        -0.048951797,
        0.038430486,
        0.023085324,
        -0.021124816,
        0.04845495,
        -0.035281092,
        0.051754724,
        -0.021783417,
        0.05185984,
        0.0013115241,
        0.023852345
      ]
    },
    {
      "id": "worldsfair-speaker-121",
      "conference": "worldsfair",
      "name": "Aaron Stanley",
      "role": "Head of Security and IT",
      "company": "dbt Labs",
      "twitter": "",
      "talks": [
        {
          "title": "AI’s Jurassic Park Period",
          "description": "Early in my career, I accidentally and unrecoverably changed data I was collecting for a federal investigation. Twenty years later, with the help of AI and a career’s worth of experience as a security leader, I intentionally did the same thing. Make no mistake, what my agent and I did together was dangerous. It was only because I had enough subject matter expertise in both the functional and risk issues that I could navigate it safely. We are in AI’s Jurassic Park period: no matter how clearly we define the rules, models will search for paths to completion. And they are very good at making those paths look safe, reasonable, and correct even when they violate policy or basic intuition. Designing the right control set is about allowing for the right expertise to be injected at the right time in the co-creation process so we can move quickly and safely into the next evolution.",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.027080124,
        0.008688276,
        0.024453064,
        -0.05437257,
        0.042347547,
        -0.01872064,
        -0.061881695,
        -0.021588512,
        0.07209696,
        -0.2363876,
        -0.028408319,
        -0.06894496,
        -0.012649487,
        0.03070075,
        -0.02147254,
        0.016555984,
        0.13527341,
        -0.016626393,
        -0.108781494,
        -0.0387263,
        -0.062488843,
        -0.081532806,
        -0.019839799,
        0.057715815,
        -0.050991908,
        0.16863383,
        0.037513882,
        0.102789685,
        -0.05805502,
        0.4473024,
        -0.10733842,
        0.032383952,
        0.041198,
        0.004570968,
        0.030333761,
        -0.11532742,
        -0.0036090158,
        -0.02762031,
        -0.028254034,
        0.03992018,
        -0.018024193,
        0.020873476,
        -0.015140139,
        0.019184358,
        -0.0052590976,
        0.018279534,
        -0.042258773,
        -0.008265436,
        0.01452323,
        0.013044846,
        -0.009074647,
        -0.014286371,
        0.06491631,
        0.08548394,
        0.0140614705,
        0.04277966,
        0.03884295,
        -0.03050021,
        0.114255555,
        -0.085777335,
        -0.096975684,
        0.021593358,
        -0.013732695,
        0.0032559677,
        0.042225387,
        -0.1002694,
        0.08276332,
        -0.026237506,
        0.030929532,
        -0.042873,
        0.04327134,
        0.087457485,
        -0.10653395,
        -0.008176907,
        -0.09942519,
        -0.07913077,
        -0.045030106,
        0.087898284,
        -0.02015017,
        0.0059043276,
        0.08639464,
        0.082859345,
        -0.02548096,
        -0.05060251,
        0.039502654,
        0.016557738,
        -0.07121737,
        0.052000858,
        0.011576566,
        0.0922809,
        -0.02768549,
        -0.07087549,
        0.04343398,
        0.07126095,
        -0.07158695,
        0.041770533,
        -0.04780662,
        0.039644334,
        -0.023019005,
        -0.0029938775,
        0.026001541,
        -0.012415768,
        0.049854517,
        -0.019341346,
        -0.03456994,
        -0.028053401,
        0.01117555,
        0.0021274122,
        0.06636819,
        -0.0023930531,
        -0.038550653,
        -0.59441906,
        0.0090946,
        -0.03770387,
        -0.048807792,
        -0.13200586,
        -0.0015181126,
        0.0076147094,
        0.07497406,
        0.049925387,
        -0.06262899,
        0.034134835,
        -0.03153365,
        -0.029605702,
        -0.0074829236,
        0.053141378,
        0.10561608,
        -0.030130627
      ]
    },
    {
      "id": "worldsfair-speaker-122",
      "conference": "worldsfair",
      "name": "Midam Kim",
      "role": "Senior Linguist and ML Engineer",
      "company": "ServiceNow",
      "twitter": "",
      "talks": [
        {
          "title": "\"My name is... my name is...\": A Linguistic Framework for Debugging Voice AI Failures",
          "description": "Every voice AI engineer has heard it: a caller repeating their name three times, getting more frustrated with each attempt. The logs look clean. Confidence scores look fine. Linguistics can help solving the mystery. By the end of this talk, you'll have a diagnostic framework for the failures that slip past standard metrics, a way to turn \"the agent just didn't get it\" into concrete, debuggable failure modes. The framework maps three levels of linguistic structure (sounds, words, and interactions) against the two dimensions every voice agent engineer already works in: what we hear (speech recognition) and what we speak (speech synthesis). That 3×2 grid surfaces problems your current tooling can't see, including: 1. Why your user cannot make your system understand their name 2. Why a single well-intentioned vocabulary hint can cause catastrophic drops in a non-English language 3. Why a transcript that's \"cumulatively correct\" can still ruin the user experience Drawing on examples from production multilingual voice AI work, I'll show where linguistic expertise connects to the engineering decisions you're already making and where it reveals failure modes that confidence scores will never warn you about. Who this is for: Voice AI engineers, ML practitioners on Voice AI pipelines, and anyone who's watched clean logs while their agent quietly fails real users.",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.028832758,
        0.04487633,
        0.0071095335,
        0.03183905,
        -0.046779897,
        -0.013813543,
        -0.065712504,
        0.0247723,
        0.072825655,
        -0.26220247,
        -0.071023405,
        -0.016941195,
        0.035743322,
        0.005766235,
        -0.03897839,
        0.042709537,
        0.05940709,
        -0.09813811,
        -0.02195588,
        -0.057270017,
        -0.05157936,
        -0.010601057,
        0.027920062,
        0.079996385,
        -0.04293681,
        0.05304989,
        0.06272753,
        0.08327982,
        -0.06991582,
        0.43249714,
        -0.14649552,
        0.0023721738,
        -0.0063098418,
        -0.06051773,
        0.027538331,
        -0.12213422,
        -0.0047895564,
        -0.021147823,
        0.0076207486,
        0.083970055,
        -0.055127613,
        0.071769096,
        -0.028676888,
        0.043519694,
        0.0021656975,
        0.0061344565,
        -0.07344228,
        0.06528566,
        -0.011958352,
        -0.019919453,
        -0.03495942,
        -0.020485986,
        0.037753314,
        0.0011936873,
        0.005508154,
        0.029774174,
        0.041015483,
        -0.041065395,
        0.1188515,
        -0.021500077,
        -0.09598566,
        0.06899123,
        0.017854514,
        0.019419722,
        0.023283059,
        -0.063339174,
        -0.007102807,
        -0.03881404,
        -0.074686065,
        -0.0027045486,
        0.06903945,
        -0.014399593,
        -0.080179855,
        -0.026808882,
        -0.028426874,
        -0.09185576,
        -0.015044754,
        0.066282764,
        -0.042003993,
        -0.025520831,
        0.041007474,
        0.05942817,
        -0.087240994,
        -0.061942164,
        0.0477867,
        -0.08059071,
        -0.06450424,
        0.029122896,
        -0.04496976,
        0.059445865,
        0.031873357,
        -0.013886028,
        -0.032668367,
        -0.009072765,
        -0.100598946,
        0.010787934,
        -0.04559154,
        0.05518436,
        -0.075125866,
        0.08266709,
        -0.029826399,
        -0.03357864,
        0.027429854,
        -0.03715249,
        -0.013788129,
        -0.0031959012,
        -0.07725451,
        -0.07047384,
        0.066324,
        0.020971265,
        -0.09864383,
        -0.61828125,
        0.051970158,
        -0.028132163,
        -0.011171864,
        -0.14080752,
        0.022554992,
        -0.023971988,
        0.04239405,
        0.061316323,
        -0.0313304,
        0.052078746,
        -0.027466893,
        -0.051277027,
        -0.039724797,
        0.05675525,
        0.0616214,
        0.005820173
      ]
    },
    {
      "id": "worldsfair-speaker-123",
      "conference": "worldsfair",
      "name": "Eyal Blum",
      "role": "",
      "company": "Figma",
      "twitter": "",
      "talks": [
        {
          "title": "How to Get Your Org to Adopt Coding Agents (Without Shipping Garbage)",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0035257163,
        -0.07358386,
        -0.0280222,
        -0.034350388,
        -0.03650103,
        0.010329049,
        -0.06456467,
        -0.015701387,
        0.005426676,
        -0.18458758,
        -0.07457323,
        -0.011821144,
        0.059513748,
        0.014482836,
        -0.02714767,
        0.004898615,
        0.11021516,
        -0.08300841,
        -0.035069294,
        -0.094652705,
        -0.04310585,
        -0.02593411,
        0.012449376,
        0.06502741,
        0.017089449,
        0.06366323,
        -0.039273147,
        0.063841894,
        -0.15065525,
        0.5107095,
        -0.051840287,
        -0.0074435407,
        0.041044552,
        -0.04233149,
        0.008539741,
        -0.058347438,
        -0.06417764,
        0.026601363,
        0.014117203,
        0.04906286,
        -0.061544344,
        0.031669635,
        -0.030977983,
        0.027293459,
        -0.062983185,
        -0.042440455,
        -0.0614249,
        -0.02171176,
        -0.051660344,
        0.007308269,
        -0.034006007,
        0.016386434,
        0.009775843,
        -0.06904974,
        0.027547162,
        0.012876658,
        0.03706267,
        -0.08268269,
        0.0786913,
        0.0049872445,
        -0.09194165,
        -0.016219962,
        -0.017837292,
        -0.009634546,
        0.03338246,
        -0.0058612563,
        0.026120543,
        -0.10945031,
        0.007669049,
        -0.06185755,
        0.00584176,
        0.07313313,
        -0.039301388,
        0.02106979,
        -0.07868065,
        -0.08767007,
        0.0073248893,
        0.10596736,
        -0.002893016,
        -0.023823677,
        0.014961168,
        0.037329283,
        -0.13724016,
        -0.00008661308,
        0.05341744,
        -0.028413659,
        -0.04916875,
        0.046496842,
        0.029695319,
        0.029498518,
        0.06569823,
        -0.04994921,
        -0.003947822,
        0.032768004,
        -0.09606983,
        0.02272617,
        -0.095805444,
        0.018014261,
        -0.01853304,
        0.021555776,
        -0.044923745,
        -0.036518957,
        0.023173144,
        -0.003929858,
        -0.0029193661,
        -0.06295645,
        0.015702635,
        0.033202183,
        0.044270433,
        -0.03972764,
        -0.01088274,
        -0.6098858,
        -0.03868169,
        -0.042832665,
        -0.034282606,
        -0.17318806,
        -0.008276098,
        -0.012712371,
        -0.024771584,
        0.09336722,
        -0.041449852,
        0.059373606,
        0.007913363,
        -0.0139628425,
        -0.0002660445,
        0.039486308,
        0.022361195,
        -0.004144729
      ]
    },
    {
      "id": "worldsfair-speaker-124",
      "conference": "worldsfair",
      "name": "Dan Bjornn",
      "role": "",
      "company": "Lease End",
      "twitter": "",
      "talks": [
        {
          "title": "Your Fine-Tuned Model Is Tech Debt: A 50x ROI House of Cards",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03549236,
        0.0059169466,
        0.09066954,
        0.014195609,
        0.029143812,
        -0.044596158,
        -0.048991755,
        0.008756522,
        0.06499503,
        -0.21451858,
        -0.08659971,
        -0.023840895,
        0.0009851876,
        0.044759803,
        -0.08619457,
        0.03050439,
        0.09922481,
        -0.095174216,
        -0.046408657,
        -0.057794858,
        -0.043060265,
        -0.084797114,
        -0.019716047,
        -0.004295467,
        -0.018400965,
        0.03495015,
        -0.029690258,
        0.07893909,
        -0.1226888,
        0.4628772,
        -0.0791118,
        -0.0035924725,
        0.0024578457,
        0.005224234,
        0.002873115,
        -0.123797365,
        0.0031944327,
        0.0844161,
        0.043045137,
        0.08507869,
        -0.030127805,
        -0.0056017167,
        -0.025387518,
        0.00607867,
        -0.001946146,
        0.0399519,
        -0.06407267,
        -0.05980005,
        0.010956115,
        -0.013577892,
        -0.06443536,
        -0.0331453,
        0.010097637,
        -0.042410463,
        0.020891057,
        0.0035295114,
        0.011288792,
        -0.06494182,
        0.09270897,
        -0.05167944,
        -0.08348525,
        0.01410977,
        -0.049896702,
        0.0019239304,
        0.03664131,
        -0.021629171,
        0.049422555,
        -0.05851858,
        0.023809629,
        -0.00851765,
        0.06414703,
        0.03115667,
        -0.06280384,
        0.033064235,
        -0.03531751,
        0.018009258,
        0.00055890874,
        0.09181056,
        -0.09398997,
        0.032368626,
        0.059285063,
        0.019902658,
        -0.12716249,
        -0.07607609,
        0.06325564,
        -0.05525105,
        -0.052050784,
        0.09482801,
        -0.036405023,
        0.111596994,
        0.0169166,
        -0.014824242,
        -0.013559741,
        -0.0144966245,
        -0.09004768,
        0.03168778,
        -0.043426234,
        0.036648497,
        -0.047695514,
        -0.022047883,
        -0.07137369,
        -0.042671762,
        0.026726574,
        -0.016312532,
        -0.06727463,
        -0.113123536,
        0.0015497218,
        -0.027108427,
        0.048709955,
        -0.072575755,
        -0.018358907,
        -0.6113685,
        -0.03862748,
        -0.03921985,
        -0.00009779822,
        -0.14544445,
        0.015460386,
        0.0046960874,
        0.034688875,
        0.050640386,
        -0.019809004,
        -0.0016783491,
        -0.019932829,
        -0.014609693,
        -0.002886247,
        0.09459309,
        0.0624878,
        0.0018700877
      ]
    },
    {
      "id": "worldsfair-speaker-125",
      "conference": "worldsfair",
      "name": "Dex Horthy",
      "role": "Founder & CEO",
      "company": "HumanLayer",
      "twitter": "https://x.com/dexhorthy",
      "talks": [
        {
          "title": "Scaling coding agents across an entire team",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.01310256,
        0.07246721,
        0.02853757,
        0.030011212,
        0.01668192,
        -0.00095120503,
        -0.018817931,
        0.0298533,
        0.055883236,
        -0.17886366,
        -0.02237324,
        -0.05073784,
        -0.059256848,
        -0.019744674,
        -0.108498216,
        -0.028690472,
        0.13456522,
        -0.0980927,
        -0.060263094,
        -0.10088869,
        -0.08818226,
        -0.06533834,
        -0.050146453,
        0.04461416,
        -0.03483547,
        0.11227611,
        -0.0089596845,
        0.08688279,
        -0.11710704,
        0.47057542,
        -0.074648984,
        0.012926864,
        0.052230913,
        -0.0027100046,
        0.046528455,
        -0.0686896,
        0.025881378,
        -0.029049069,
        0.039531816,
        0.082107976,
        -0.0008286734,
        0.04756553,
        -0.06343836,
        0.0075878305,
        -0.08495205,
        0.009517693,
        -0.07966133,
        -0.042365454,
        -0.054075632,
        0.0066965655,
        -0.09154423,
        0.048679356,
        0.019530073,
        -0.058402408,
        0.021058336,
        0.046274584,
        0.018441817,
        -0.13402042,
        0.04637097,
        -0.049012464,
        -0.1204576,
        -0.011173887,
        0.0012185072,
        0.01746119,
        -0.040576775,
        -0.09353029,
        0.053906724,
        -0.056225512,
        -0.0014208888,
        -0.08689062,
        0.043820858,
        0.052373044,
        -0.06603496,
        0.025790295,
        -0.054085877,
        -0.023843594,
        0.022946848,
        0.08703524,
        0.00844617,
        -0.029420966,
        -0.024847262,
        0.03874598,
        -0.04949503,
        -0.0747842,
        0.061404053,
        -0.004350434,
        -0.001025442,
        -0.004069883,
        0.0075538326,
        0.032956243,
        0.030694665,
        -0.03858522,
        -0.026190095,
        0.01339944,
        -0.05259844,
        -0.0013205959,
        -0.12564483,
        0.033784043,
        0.012030198,
        -0.0015561268,
        0.020502353,
        -0.07579834,
        0.037320547,
        -0.06533283,
        -0.035028048,
        0.017064713,
        0.037821677,
        -0.043656744,
        0.06301944,
        -0.039106518,
        -0.054453302,
        -0.595374,
        -0.014574981,
        -0.009358661,
        -0.062247124,
        -0.15539588,
        -0.026383664,
        0.01651223,
        0.045046248,
        0.04348364,
        -0.06635934,
        0.08557563,
        -0.08136659,
        0.008390417,
        0.025010401,
        0.020081526,
        0.017465828,
        0.0054278523
      ]
    },
    {
      "id": "worldsfair-speaker-126",
      "conference": "worldsfair",
      "name": "Sam Bhagwat",
      "role": "",
      "company": "Mastra",
      "twitter": "https://x.com/calcsam",
      "talks": [
        {
          "title": "Every Harness Will Become A Claw",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0032846977,
        0.039948538,
        0.09594686,
        -0.0034131394,
        0.046736054,
        -0.0357768,
        0.004357545,
        -0.0037801196,
        0.038098298,
        -0.2140843,
        -0.038628507,
        -0.023395091,
        0.03322525,
        0.012795432,
        -0.082614005,
        -0.007494327,
        0.10216908,
        -0.077592865,
        -0.074355185,
        -0.042748116,
        -0.012495345,
        -0.09581793,
        -0.019939868,
        -0.02169052,
        -0.057682764,
        0.00046207142,
        -0.03720131,
        0.0020138142,
        -0.104017965,
        0.47925225,
        -0.050517764,
        0.04700533,
        -0.007069747,
        -0.09368959,
        0.066464685,
        -0.07966101,
        0.041005056,
        0.012643864,
        -0.038561445,
        0.049402576,
        -0.07396545,
        0.05992612,
        -0.055777628,
        0.0012250594,
        0.006877772,
        0.029704187,
        -0.075840615,
        -0.086440325,
        0.017477354,
        -0.10271673,
        -0.03523084,
        -0.0022020545,
        0.10795262,
        -0.017231565,
        0.061159264,
        0.0437506,
        -0.019641086,
        -0.057226922,
        0.10232198,
        0.027574498,
        -0.06972492,
        0.01232417,
        -0.017570568,
        0.016202182,
        -0.008815577,
        -0.077136315,
        0.10028234,
        -0.045091767,
        -0.017584465,
        0.04075666,
        0.0180018,
        -0.0062214476,
        0.00845857,
        0.09138925,
        -0.043662537,
        -0.034042228,
        -0.04689147,
        0.09723855,
        -0.07508297,
        0.026411446,
        0.06894453,
        0.014050556,
        -0.07625239,
        -0.032137923,
        0.01650846,
        -0.01925568,
        -0.047719467,
        0.010362813,
        0.015223752,
        0.032795105,
        0.04973063,
        -0.06961578,
        0.022739055,
        -0.04579021,
        -0.04894028,
        0.057798654,
        -0.09387605,
        0.03800683,
        -0.05677524,
        -0.008861233,
        -0.028667303,
        -0.007897713,
        0.05525499,
        0.005062838,
        -0.013060769,
        -0.04169315,
        0.00031741618,
        -0.078981094,
        0.081332535,
        -0.04527089,
        0.010548666,
        -0.6213235,
        -0.0046816226,
        -0.0076079085,
        -0.0057297186,
        -0.12698023,
        -0.025996141,
        -0.026792476,
        0.031558495,
        0.095881194,
        -0.012000191,
        0.064824775,
        -0.041035213,
        0.039072916,
        0.0051396363,
        0.019161949,
        0.022816513,
        0.06185499
      ]
    },
    {
      "id": "worldsfair-speaker-127",
      "conference": "worldsfair",
      "name": "Armen Aghajanyan",
      "role": "Co-Founder & CEO",
      "company": "Perceptron AI",
      "twitter": "",
      "talks": [
        {
          "title": "Perceptron Mk1 — Perceptron Inc",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.060707998,
        0.0155264195,
        -0.021773348,
        -0.012186549,
        0.0009907547,
        0.03817316,
        -0.024203107,
        0.035331532,
        0.089761995,
        -0.22179902,
        -0.013180589,
        -0.017050315,
        0.028731095,
        -0.014163635,
        -0.07300819,
        -0.08612791,
        0.07576014,
        -0.09560712,
        -0.032446504,
        -0.10659618,
        -0.0102995,
        -0.060577147,
        0.008488843,
        0.053326465,
        -0.07919719,
        0.07479268,
        0.047134776,
        0.06793367,
        -0.09876829,
        0.48162442,
        -0.05853932,
        -0.056352135,
        0.010656637,
        -0.025969936,
        0.043728884,
        -0.091460176,
        -0.013542663,
        -0.037457574,
        -0.0040515047,
        0.06160693,
        -0.06332438,
        0.0222301,
        -0.013554817,
        0.036498826,
        0.032530542,
        -0.02387609,
        -0.123151734,
        -0.054239716,
        0.033487048,
        -0.026108071,
        -0.07749639,
        -0.006897222,
        0.042716425,
        0.00004826615,
        0.06865233,
        0.019982714,
        0.037157204,
        -0.037721775,
        0.06083951,
        -0.1194779,
        -0.07207218,
        -0.008026328,
        0.0015491613,
        0.057811774,
        -0.048640188,
        -0.0065222518,
        0.07667141,
        -0.06268235,
        0.0086098835,
        0.0040869284,
        -0.04092107,
        0.058607537,
        -0.07800672,
        0.030827904,
        -0.027157042,
        0.02738921,
        -0.0027857006,
        0.042385645,
        0.035582673,
        -0.032832894,
        0.024242543,
        0.018554471,
        -0.023949012,
        -0.044821158,
        0.05326933,
        -0.018763093,
        -0.09315452,
        0.07439369,
        0.01753077,
        -0.0003947643,
        0.09793211,
        -0.04769278,
        -0.02319821,
        -0.018986268,
        -0.018065762,
        0.041032273,
        -0.091480605,
        0.024053955,
        0.059532773,
        0.020749498,
        0.00161368,
        -0.04607239,
        0.012979192,
        0.006977007,
        -0.026317904,
        -0.048576284,
        0.014670703,
        -0.015426555,
        0.03657439,
        -0.026501814,
        0.0005012204,
        -0.6269673,
        0.0074772416,
        -0.004070949,
        -0.012489948,
        -0.15623806,
        -0.053758614,
        -0.038893696,
        0.06877501,
        0.05184693,
        -0.034832694,
        0.014830326,
        -0.051667884,
        0.104243115,
        0.0008731665,
        -0.00979311,
        -0.041828394,
        -0.0014582998
      ]
    },
    {
      "id": "worldsfair-speaker-128",
      "conference": "worldsfair",
      "name": "Fouad Matin",
      "role": "Cyber Team Lead",
      "company": "OpenAI",
      "twitter": "https://x.com/fouadmatin",
      "talks": [
        {
          "title": "OpenAI Cyber talk",
          "description": "",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.014861755,
        -0.046918724,
        0.0031548014,
        0.017782535,
        0.0601241,
        0.007799615,
        -0.024721704,
        0.024005722,
        0.042484626,
        -0.23831296,
        -0.0032293547,
        -0.04840666,
        -0.03135614,
        -0.04466014,
        -0.048035815,
        -0.036868785,
        0.08058863,
        -0.011816833,
        -0.03642376,
        -0.021194976,
        -0.08006641,
        -0.107790545,
        0.08909615,
        0.016994165,
        -0.07384453,
        0.044222225,
        0.030869272,
        0.062114622,
        -0.12700626,
        0.452523,
        -0.08788618,
        -0.0280665,
        0.06522529,
        -0.004942809,
        0.029654365,
        -0.08922742,
        -0.04328752,
        -0.038597986,
        0.0013227714,
        0.0854043,
        -0.033489294,
        0.05419014,
        -0.06457075,
        0.08221944,
        0.006286835,
        0.009818163,
        -0.07739352,
        -0.026969856,
        -0.037516322,
        0.012246478,
        -0.04360019,
        -0.0057930336,
        -0.033507522,
        0.036567107,
        0.056784835,
        0.04966043,
        -0.03266489,
        -0.0478544,
        0.098402165,
        -0.09589802,
        -0.091250755,
        0.035716034,
        0.0010890858,
        0.0013907461,
        0.05652665,
        -0.074396625,
        0.09653336,
        -0.0726769,
        -0.06656654,
        0.03440488,
        0.061569788,
        0.030485569,
        -0.00908416,
        0.027213296,
        -0.09922027,
        -0.067627534,
        -0.014033656,
        0.06375227,
        -0.01812739,
        -0.042018548,
        0.06515621,
        0.06320127,
        -0.08580092,
        -0.037606318,
        0.02957141,
        -0.07781002,
        -0.014013747,
        0.090477936,
        -0.04806434,
        0.007906296,
        0.06112437,
        -0.058808453,
        -0.03594996,
        0.030644506,
        -0.06021007,
        0.024971288,
        -0.1051868,
        0.04423078,
        -0.044631727,
        -0.025836445,
        0.04975765,
        -0.09846713,
        0.062021963,
        -0.0009381717,
        0.0077149183,
        -0.010038797,
        0.03297604,
        0.005133338,
        0.08790217,
        0.031451017,
        -0.097412385,
        -0.603751,
        0.025055874,
        -0.04771932,
        0.019211508,
        -0.16205448,
        -0.012010596,
        -0.025357608,
        0.026503948,
        0.07738475,
        -0.02732526,
        0.018306768,
        -0.0412929,
        -0.00028872053,
        0.009441603,
        0.009562455,
        -0.004613414,
        -0.0070073237
      ]
    },
    {
      "id": "worldsfair-speaker-129",
      "conference": "worldsfair",
      "name": "Amit Desai",
      "role": "Director of Voice AI",
      "company": "Roku",
      "twitter": "",
      "talks": [
        {
          "title": "The Goldilocks problem: when your Robot asks too much — or acts too soon.",
          "description": "Embodied agents are crossing from answering questions to taking physical actions — moving a box, turning a wheel — and people will command them by voice, because voice is the fastest, most natural interface we have. But voice is also the most error-prone, and when a misheard command drives a physical action, the failure isn't a wrong answer; it's human harm, damage, or an expensive, irreversible mistake. The field has never needed a serious way to handle voice-command errors, because informational agents made them cheap. Embodiment ends that. This talk replaces the usual hand-waving — \"don't ask too much, don't get it wrong too much\" — with a single number you can optimize. The core idea: both confirming and erring cost the user. A confirmation is friction — attention, time, a delayed action; a wrong action is a mistake cost, often higher given physical harm or expense. Put them on one ledger and you can measure a voice interface as average user cost per command, and make minimizing it the system's objective. From that falls a non-obvious rule — you confirm or not based on both cost and uncertainty: an expected value. I'll frame confirmation as just one option alongside acting, disambiguation (choices), and deferring; reason at the level of goals rather than low-level motion; walk the architecture (task hypotheses → user-cost model → confirmation policy); and show eval results from a simulated environment measuring regret against oracle behavior. I'll close with what worked applying this to voice in smart TVs, speakers, and navigation — and a challenge to bring this metric to robots, cars, and wearables before the errors do.",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.012896799,
        0.04214504,
        -0.039067492,
        -0.008154351,
        0.021385308,
        -0.0018982892,
        -0.11226849,
        0.016340422,
        0.034917783,
        -0.20920457,
        0.0021256893,
        -0.13666417,
        0.037111614,
        -0.043995366,
        -0.05102185,
        -0.016111938,
        0.08168953,
        -0.037738018,
        -0.04359625,
        -0.02000206,
        0.004923654,
        -0.033517376,
        -0.022733727,
        0.08062526,
        -0.027176093,
        0.054373045,
        0.048055958,
        0.08276553,
        -0.10148473,
        0.414317,
        -0.13408329,
        -0.047359295,
        -0.000032409684,
        0.0042216536,
        0.07036959,
        -0.08869383,
        0.0114119025,
        -0.038002428,
        0.0194332,
        0.028342448,
        0.00065886637,
        0.05675871,
        -0.074132584,
        0.038943134,
        0.0566237,
        -0.0737858,
        -0.07267367,
        0.017892465,
        -0.003315671,
        -0.005442224,
        -0.08517522,
        0.0038971647,
        -0.03259308,
        -0.059970696,
        0.020979863,
        0.05389987,
        0.03498654,
        -0.090666376,
        0.109690405,
        -0.032610416,
        -0.12035916,
        0.029066093,
        0.04516128,
        0.028317044,
        0.007481451,
        -0.07439075,
        0.053895462,
        -0.09354571,
        -0.0063212053,
        0.014195186,
        0.07300794,
        0.03608489,
        -0.11608289,
        0.014228752,
        -0.083385356,
        -0.10118564,
        -0.04691085,
        0.060036395,
        -0.00045016466,
        0.009102564,
        0.02267569,
        0.008048872,
        -0.047866456,
        -0.0010196591,
        0.070189565,
        -0.058141015,
        -0.024999402,
        0.07606338,
        0.008600564,
        0.00395389,
        0.022592252,
        -0.015073363,
        -0.08421161,
        -0.01563384,
        -0.10891684,
        -0.032133356,
        -0.07564833,
        0.017522896,
        -0.022554109,
        0.01797804,
        -0.025698323,
        0.0028457465,
        0.02232505,
        0.05882894,
        -0.020316152,
        -0.023394441,
        0.04448239,
        0.0023918445,
        0.082646295,
        -0.050145518,
        -0.053449355,
        -0.6379279,
        -0.0074810456,
        -0.028953489,
        0.003386825,
        -0.16037853,
        0.063694485,
        -0.039259233,
        0.05690102,
        0.039996218,
        -0.012603698,
        0.05602536,
        -0.033749595,
        -0.028408216,
        0.027848756,
        0.052565634,
        0.06309785,
        -0.025487877
      ]
    },
    {
      "id": "worldsfair-speaker-130",
      "conference": "worldsfair",
      "name": "Connie Fan",
      "role": "PM Lead",
      "company": "Google DeepMind",
      "twitter": "https://x.com/aurielws",
      "talks": [
        {
          "title": "How to Stop Shipping Low-Quality RL Environments",
          "description": "Training harnesses are data generators: when environments are flaky, stale, reward-hacked, or mismatched to production, models learn the wrong behavior. This talk distills common RL environment failures across coding, SaaS, and support-agent workflows, then offers a practical framework for building production-grade harnesses with clean signal, realistic state, fail-fast behavior, and trustworthy rewards.",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.084265485,
        0.045783024,
        -0.0018477826,
        -0.024328347,
        0.024217399,
        -0.047670476,
        -0.077901125,
        -0.0016323504,
        0.0476541,
        -0.24608044,
        -0.009406345,
        -0.068109125,
        -0.021695424,
        -0.009676651,
        -0.089320995,
        0.04115093,
        0.08350367,
        -0.05279149,
        -0.0661458,
        -0.06114384,
        -0.022849048,
        -0.037586357,
        0.04308146,
        0.074458316,
        -0.022672024,
        0.03649693,
        0.018962234,
        0.033579696,
        -0.09765775,
        0.43235213,
        -0.12134369,
        0.008667712,
        -0.0066381744,
        -0.034676775,
        0.06782013,
        -0.11149181,
        -0.025962804,
        0.0010012378,
        0.030033084,
        0.07865229,
        -0.03633582,
        0.06958407,
        -0.018586542,
        0.02006494,
        -0.067390606,
        -0.06857245,
        -0.089571565,
        0.012868889,
        -0.012988011,
        0.026221566,
        -0.07888902,
        -0.001424687,
        -0.0036603657,
        -0.030799957,
        0.037153482,
        0.06375127,
        -0.03603671,
        -0.01120903,
        0.045113772,
        0.03507683,
        -0.111970246,
        0.044117242,
        -0.02185619,
        -0.02926319,
        0.0016258133,
        -0.07639356,
        0.048174053,
        -0.08565705,
        0.014067612,
        -0.023332385,
        0.060779557,
        -0.010046651,
        -0.034384593,
        0.021124722,
        -0.09432474,
        -0.107310705,
        -0.0045766127,
        0.07318554,
        -0.048299734,
        -0.020348018,
        0.029875051,
        0.09653316,
        -0.06810239,
        -0.06971174,
        0.05787378,
        0.013359545,
        -0.0926727,
        0.021653729,
        -0.0683562,
        0.07933682,
        0.05135583,
        -0.023770636,
        -0.06038405,
        -0.03751971,
        -0.07517367,
        0.012900437,
        -0.07839044,
        0.042786967,
        -0.07458718,
        0.04836744,
        0.00599093,
        0.034885745,
        0.031152729,
        0.039440546,
        -0.090477854,
        -0.026867356,
        -0.027465533,
        -0.025645189,
        0.040798523,
        0.0026314368,
        -0.048236288,
        -0.61736417,
        -0.031475756,
        -0.030319888,
        -0.032067392,
        -0.16123375,
        0.015450432,
        -0.033995338,
        -0.0016639131,
        0.0945015,
        -0.0031080656,
        0.041656155,
        -0.011895754,
        -0.01747385,
        0.001882999,
        0.05535693,
        0.077613235,
        0.06410307
      ]
    },
    {
      "id": "worldsfair-speaker-131",
      "conference": "worldsfair",
      "name": "Oscar Mullin",
      "role": "SVP of Technology",
      "company": "MercadoLibre",
      "twitter": "",
      "talks": [
        {
          "title": "AI-transforming 18K engineers, 40K repos, and an agent swarm: what worked, what didn't",
          "description": "Most AI transformation talks open with a timeline of how coding has evolved. We promise we won't. We **doubled delivery throughput** across **18K engineers** and **40K repos**, **improved performance with auto-research**, and **migrated 9,000 apps with autonomous agents**. We also built things that became obsolete in three months, picked the wrong abstraction at least twice, and have a graveyard of internal tools we'd rather forget. You'll get the architecture, the metrics that actually moved, and the ones we wish we hadn't measured. Still in progress. Already worth talking about. 18K engineers, **swarm of agents** and **thousands of new builders who don't code**.",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.016520489,
        0.0790992,
        -0.028942835,
        0.041986335,
        -0.042617176,
        -0.030489163,
        -0.101794645,
        0.0071437545,
        0.045213223,
        -0.19238809,
        -0.004605013,
        -0.015097641,
        0.00352712,
        -0.036903773,
        -0.040620755,
        0.008644944,
        0.13062456,
        -0.08482471,
        -0.050534394,
        -0.059773047,
        -0.01811243,
        -0.09394425,
        -0.007460611,
        0.06922774,
        -0.023796488,
        0.090670496,
        -0.0009818351,
        0.07143359,
        -0.11024813,
        0.48175678,
        -0.12174045,
        0.016505543,
        0.04203068,
        -0.051719815,
        0.10079024,
        -0.05713039,
        -0.007177958,
        -0.007903783,
        -0.041470844,
        0.0134773515,
        -0.06030227,
        0.06770445,
        -0.068981014,
        0.061522346,
        -0.011784213,
        -0.070417,
        0.008312584,
        0.012867491,
        0.015118572,
        -0.056849238,
        -0.03603836,
        -0.019790843,
        0.0011510076,
        0.0041753184,
        0.110349074,
        -0.04419844,
        0.00005069157,
        -0.06617241,
        0.0807415,
        -0.07856625,
        -0.0846033,
        -0.009746052,
        0.060993716,
        0.010107864,
        -0.030016888,
        -0.15102507,
        0.07070828,
        -0.06741339,
        0.009234164,
        -0.04531528,
        -0.028386703,
        0.065942004,
        -0.06800608,
        -0.0022287946,
        -0.045714367,
        -0.142755,
        -0.044796325,
        0.11381302,
        -0.06323041,
        -0.029895185,
        -0.00045456842,
        0.008497105,
        -0.08598181,
        0.0009442251,
        0.057901006,
        0.06637397,
        -0.093741015,
        0.032771014,
        -0.02642372,
        0.05399719,
        -0.0039989245,
        0.011692922,
        -0.022543615,
        0.0033331027,
        -0.027712144,
        0.012369239,
        -0.044321056,
        0.054493256,
        -0.013931103,
        0.08281408,
        -0.013787093,
        -0.06612589,
        0.05233606,
        0.01331779,
        0.01173372,
        -0.040328253,
        -0.035237692,
        0.016570715,
        0.061931714,
        -0.056831446,
        0.018507006,
        -0.57948756,
        0.023261707,
        -0.01191149,
        -0.007948013,
        -0.1364907,
        0.03305752,
        0.002728339,
        0.032313336,
        0.09401033,
        0.016274117,
        0.066972926,
        -0.06864669,
        -0.00084455573,
        -0.039207086,
        0.010825074,
        0.046377823,
        0.029890325
      ]
    },
    {
      "id": "worldsfair-speaker-132",
      "conference": "worldsfair",
      "name": "Tushar Jain",
      "role": "EVP of Engineering",
      "company": "Docker",
      "twitter": "",
      "talks": [
        {
          "title": "Unlock Agent Autonomy: The Runtime for AI-Native Systems",
          "description": "The way software gets built in 2026 doesn't look like it did in 2024. The actors changed. Agents read and write entire codebases. Subagents spawn to chase down a flaky test, refactor a module, or triage an incident. But this shift doesn't stop at the SDLC. Agents increasingly invoke tools, interact with enterprise systems, install dependencies, call APIs, and orchestrate workflows across local machines, CI systems, cloud infrastructure, and organizational boundaries. The teams leaning into this shift are moving faster, and the gap is widening by the quarter.\n\nBut few have the confidence to let agents operate autonomously across those environments. Not because the model capability isn't there. Trust isn't. Agents can pull a poisoned dependency, invoke an untrusted tool, wipe a database, leak sensitive data, or access systems they shouldn't. Prompt-level instructions won't close that gap, the unlock has to happen one layer down, at the runtime layer itself.\n\nDocker spent the last decade making it safe to ship software by getting the runtime right: isolation, network policy, trusted base images, and credentials. Agents are the next workload, and the same principles apply. Tushar Jain, EVP of Engineering at Docker, walks through what the runtime layer for AI-native systems looks like in practice: hardened runtime foundations, sandboxes that constrain what agents can touch, and governance controls that limit what agents can introduce, access, and execute across local, CI, cloud, and enterprise environments.",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.055893157,
        0.08565661,
        -0.009523175,
        0.015242601,
        0.035601355,
        -0.025438149,
        -0.12728402,
        0.04085642,
        0.042642366,
        -0.20697199,
        -0.018792834,
        -0.06664998,
        0.033743158,
        -0.044356532,
        -0.06421303,
        -0.005620066,
        0.121127084,
        -0.06345873,
        -0.071010135,
        -0.04950537,
        -0.020369323,
        -0.060687315,
        -0.031912737,
        0.045671407,
        -0.021888386,
        0.106857195,
        0.033307556,
        0.030722542,
        -0.10408449,
        0.4043904,
        -0.037710305,
        -0.0028671199,
        0.015141145,
        -0.07068876,
        -0.007435907,
        -0.001214257,
        -0.06352129,
        0.016930507,
        -0.015471597,
        0.034728415,
        0.0003934756,
        0.07654042,
        -0.057005323,
        0.04355565,
        -0.044605147,
        -0.0980619,
        -0.04899976,
        0.011340086,
        -0.015721802,
        -0.058573797,
        -0.08840123,
        -0.0044858484,
        0.10796729,
        -0.056132432,
        0.052716054,
        0.042761184,
        0.06784858,
        -0.087482415,
        0.12124068,
        -0.04848207,
        -0.17994837,
        -0.0048351362,
        0.049743377,
        0.0035590625,
        -0.029847255,
        -0.1039205,
        0.01595169,
        -0.018421326,
        -0.020267421,
        -0.037834425,
        0.018123072,
        0.038668912,
        -0.0702538,
        0.0699054,
        -0.07321213,
        -0.11342021,
        -0.06202288,
        0.12188476,
        0.025240213,
        -0.035095338,
        -0.06744741,
        0.053438533,
        -0.10518533,
        -0.06698075,
        0.051974975,
        0.016610809,
        -0.050825104,
        0.0079245735,
        -0.031332932,
        0.089674816,
        0.054080807,
        -0.013305085,
        -0.07135322,
        0.05139878,
        -0.0837961,
        0.05030603,
        -0.115547925,
        -0.0022697723,
        -0.067475885,
        0.0066097644,
        0.009435608,
        -0.07090943,
        0.06755197,
        0.018042088,
        -0.026777416,
        -0.001542583,
        0.03594008,
        -0.030660762,
        0.05249009,
        -0.013559655,
        0.016911363,
        -0.590482,
        0.024329223,
        0.0011916866,
        -0.03202021,
        -0.092144884,
        0.01999856,
        0.036338348,
        0.12521718,
        0.10737494,
        -0.018393245,
        0.072110474,
        -0.055656623,
        -0.037628263,
        -0.05774247,
        -0.0086441375,
        0.00018525544,
        0.020009372
      ]
    },
    {
      "id": "worldsfair-speaker-133",
      "conference": "worldsfair",
      "name": "Cam Kennedy",
      "role": "",
      "company": "Airbyte",
      "twitter": "",
      "talks": [
        {
          "title": "How We Built the Airbyte Agent MCP Server and CLI",
          "description": "Agents need a reliable way to reach live business data. At Airbyte we built two interfaces for that, and this session is how.\n\nCam built much of that surface. He covers the MCP server that exposes hundreds of sources through one endpoint with managed auth, and the CLI that's designed for agent harnesses rather than humans, with embedded help, packaged agent skills, and no credentials passed over the command line. Expect the real engineering: why a CLI turned out to fit autonomous agents better than the API or SDK, how auth works across the layers, and the tradeoffs the team made along the way.\n\nCome if you're building agent tooling or thinking about how to expose your own systems to agents cleanly.",
          "day": "Day 2 — Session Day 1",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.009118793,
        0.01006408,
        -0.04178689,
        0.03877111,
        0.053601723,
        -0.008106861,
        -0.09357966,
        -0.0036113714,
        0.02302359,
        -0.23336627,
        0.024376595,
        -0.07438904,
        0.067223646,
        0.011811316,
        0.032284204,
        -0.023315424,
        0.10365719,
        -0.08948646,
        -0.051610302,
        -0.06659617,
        -0.069245726,
        -0.10961092,
        -0.003644891,
        0.022794483,
        -0.037149854,
        0.06307879,
        -0.06306109,
        0.06688829,
        -0.117702596,
        0.48173386,
        -0.04334086,
        -0.013767501,
        0.0025927986,
        -0.036288258,
        0.03090001,
        -0.030786198,
        -0.06525553,
        0.06451905,
        0.0052834786,
        -0.032978456,
        0.0062692505,
        0.029163873,
        -0.08598518,
        -0.03545769,
        -0.051176425,
        0.029809514,
        -0.07245973,
        -0.066345796,
        0.004376686,
        -0.01490564,
        -0.06116906,
        0.048726965,
        0.05862506,
        0.042664386,
        0.0024209032,
        0.0129075525,
        -0.0221159,
        -0.054547697,
        0.07114545,
        -0.058221843,
        -0.05431794,
        0.011879274,
        -0.013543462,
        -0.020132056,
        0.0027646178,
        -0.02357982,
        0.06180432,
        -0.041245006,
        0.04381936,
        -0.053056713,
        0.033179525,
        0.011924252,
        -0.06292321,
        0.005066387,
        -0.11498677,
        -0.11865508,
        -0.025941575,
        0.044801235,
        0.00823176,
        0.05285755,
        0.014337364,
        0.027892485,
        -0.09330905,
        -0.11201034,
        -0.034467053,
        -0.034328163,
        -0.07044884,
        0.0010188704,
        -0.041629873,
        0.011755542,
        0.014297382,
        -0.0224047,
        0.01877903,
        -0.053879324,
        -0.040058043,
        -0.05838427,
        0.019963106,
        -0.015140365,
        -0.04476984,
        0.06795863,
        -0.07088344,
        -0.0040936903,
        0.072186604,
        0.0012254241,
        -0.05244329,
        -0.021467375,
        -0.0028261736,
        -0.04237696,
        -0.0014316165,
        -0.057796955,
        -0.0036052114,
        -0.6242531,
        -0.04021496,
        -0.0014117829,
        0.009621171,
        -0.13387309,
        0.041531816,
        -0.040304236,
        0.032593925,
        0.054903,
        -0.050012697,
        0.06960519,
        0.02433915,
        0.027266309,
        -0.016083037,
        -0.020772414,
        0.06458993,
        -0.064663246
      ]
    },
    {
      "id": "worldsfair-speaker-134",
      "conference": "worldsfair",
      "name": "Erik Meijer",
      "role": "Computer scientist and entrepreneur",
      "company": "Leibniz Labs",
      "twitter": "https://x.com/headinthebox",
      "talks": [
        {
          "title": "In Code They Act, In Proof We Trust",
          "description": "AI agents today execute on blind trust, and the failure modes are already in the headlines: a dealership chatbot agreeing to sell a $76,000 Chevy Tahoe for $1, a coding agent wiping a production database during a code freeze, and an \"agent skill\" installing a keylogger on a developer's machine. Automind enforces a different discipline: before any action runs, the agent submits an execution plan plus a machine-checkable proof of safety and correctness in Universalis, and a small checker decides whether the plan is allowed to execute. The result is left-shifted trust, with policy compliance established before the first side effect.",
          "day": "Day 2 — Session Day 1",
          "time": "4:50pm-5:10pm",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.07692891,
        0.022302553,
        0.032972578,
        0.032411993,
        0.031309903,
        -0.023835171,
        -0.08071481,
        0.02593361,
        0.062337596,
        -0.18588702,
        -0.027059814,
        -0.03987079,
        0.0048609152,
        -0.055587277,
        -0.04071972,
        -0.011809583,
        0.06462641,
        -0.09380067,
        -0.06872487,
        -0.13123606,
        -0.029843718,
        -0.055282027,
        -0.00029957457,
        0.08738517,
        -0.11901337,
        0.077663384,
        -0.007231789,
        0.06368047,
        -0.08689665,
        0.46381623,
        -0.09009823,
        0.027538758,
        0.049232338,
        -0.016929848,
        0.024121506,
        -0.08710265,
        -0.07556644,
        -0.034982104,
        0.00022251823,
        0.09442536,
        -0.028727053,
        0.062010605,
        0.012929764,
        -0.011476571,
        0.000110472,
        0.004869269,
        -0.049228348,
        -0.013486485,
        0.0083361445,
        -0.034158222,
        -0.016253512,
        -0.03521633,
        0.058205444,
        -0.010241817,
        0.09128554,
        0.042002138,
        0.041789237,
        -0.027748827,
        0.04812096,
        -0.020618955,
        -0.08501796,
        -0.085791185,
        0.027368367,
        0.1135082,
        0.029589778,
        -0.11609549,
        0.03222717,
        -0.12690148,
        0.054348163,
        -0.05824065,
        0.035185438,
        0.05488725,
        -0.09446917,
        -0.019299053,
        -0.07541687,
        -0.037275348,
        -0.050481886,
        0.06355176,
        -0.017493716,
        -0.031009676,
        -0.029612105,
        0.017674085,
        -0.07793049,
        -0.079905815,
        -0.017350927,
        0.0016966219,
        -0.101169005,
        0.013375591,
        -0.023108073,
        0.019242816,
        0.027968304,
        -0.084941685,
        -0.09086166,
        0.0021389106,
        -0.09256189,
        0.024443569,
        -0.0651598,
        0.024169784,
        0.05993305,
        0.0031581812,
        -0.014678618,
        -0.022730796,
        0.11036727,
        0.011167965,
        0.024376517,
        -0.0816012,
        0.066791736,
        -0.0015202962,
        0.047227394,
        0.013807906,
        0.013555582,
        -0.5741782,
        -0.050734397,
        0.018187325,
        -0.015792582,
        -0.15461463,
        0.038643517,
        -0.0068541244,
        0.10322291,
        0.13544291,
        0.050317395,
        0.06557423,
        -0.027645074,
        0.0389887,
        -0.017340539,
        0.0073125055,
        0.030304613,
        -0.008163322
      ]
    },
    {
      "id": "worldsfair-speaker-135",
      "conference": "worldsfair",
      "name": "Barr Yaron",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "2026 AI Engineering Survey and Arize Track Intro",
          "description": "TBD — Harbor launch keynote/session details to be finalized.",
          "day": "Day 3 — Session Day 2",
          "time": "9:00am-9:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        -0.013572106,
        0.023572708,
        0.008135178,
        0.015613744,
        0.030544795,
        -0.037543997,
        -0.1283064,
        0.048495397,
        -0.002862509,
        -0.22997418,
        0.024366196,
        -0.07571079,
        -0.0061269975,
        0.03160803,
        -0.05160314,
        0.0050484193,
        0.089974225,
        -0.0687863,
        -0.0504597,
        -0.14346464,
        -0.034332752,
        -0.04172051,
        -0.040307265,
        0.042957265,
        -0.051174924,
        0.107583955,
        0.0069101807,
        0.060874987,
        -0.16970353,
        0.44066885,
        0.017928017,
        -0.059936076,
        -0.011883756,
        -0.021881275,
        0.058655456,
        -0.03552571,
        -0.04380265,
        -0.0109619135,
        -0.0064760568,
        0.08498464,
        0.067925155,
        0.02189092,
        -0.018018147,
        0.021141527,
        -0.034121566,
        -0.049097124,
        -0.047877602,
        0.012222126,
        -0.022654923,
        0.0015549472,
        -0.017841373,
        -0.06563225,
        0.030119821,
        0.00020459961,
        0.06286174,
        0.02454442,
        -0.028364748,
        -0.044058777,
        0.08148653,
        -0.05033334,
        -0.07136194,
        0.035649598,
        -0.033192948,
        0.032378323,
        -0.019213164,
        0.018574247,
        0.051875263,
        -0.06082616,
        -0.11183858,
        -0.014440474,
        -0.04820612,
        0.060687877,
        -0.042653583,
        -0.072723806,
        -0.04700807,
        -0.036681578,
        -0.027763007,
        0.038197767,
        0.010843756,
        -0.023917146,
        -0.0059313467,
        -0.01502225,
        -0.059576645,
        -0.04802983,
        -0.011320576,
        0.019560771,
        -0.054364815,
        0.14959136,
        -0.058326133,
        0.06832282,
        0.009854743,
        -0.026041595,
        -0.006304711,
        0.072964124,
        -0.08081317,
        0.038859393,
        -0.12148163,
        0.082826525,
        -0.09747478,
        -0.026635606,
        -0.0031425392,
        0.04161176,
        0.047850188,
        -0.003070917,
        -0.012848065,
        -0.040490318,
        -0.06481248,
        0.018313665,
        0.05951573,
        -0.019873843,
        -0.099977896,
        -0.5883046,
        0.04309553,
        0.009589772,
        -0.07355359,
        -0.16135608,
        0.008736785,
        -0.04413968,
        0.06932941,
        0.054811753,
        -0.007915177,
        0.09912045,
        -0.07385191,
        0.01963828,
        -0.031867053,
        0.10827666,
        0.03802914,
        0.033170026
      ]
    },
    {
      "id": "worldsfair-speaker-136",
      "conference": "worldsfair",
      "name": "Thariq Shihipar",
      "role": "Claude Code",
      "company": "Anthropic",
      "twitter": "https://x.com/trq212",
      "talks": [
        {
          "title": "Seeing like an Agent",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "9:10am-9:30am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.0026283944,
        0.044584624,
        0.02993586,
        0.015141196,
        0.033715956,
        -0.026885537,
        -0.0745737,
        0.011304413,
        0.070546,
        -0.22495244,
        0.0004716865,
        -0.04585919,
        0.038325172,
        -0.0054897415,
        -0.022807188,
        -0.012380696,
        0.106441095,
        -0.10731263,
        -0.044370383,
        -0.10655346,
        -0.05193736,
        -0.055514954,
        -0.00518637,
        0.08196844,
        -0.06689772,
        0.10078043,
        -0.001136494,
        0.028714603,
        -0.122688204,
        0.44899908,
        -0.112316586,
        -0.00744605,
        0.00038163486,
        -0.021583049,
        0.08465782,
        -0.055361744,
        0.051589742,
        0.004232263,
        0.08639784,
        0.0032648065,
        -0.013565999,
        0.03653161,
        -0.00872509,
        0.042136554,
        0.031090872,
        0.037681185,
        -0.09896193,
        -0.017658295,
        -0.033711832,
        0.023127254,
        -0.052786343,
        -0.031359974,
        0.0045325053,
        -0.012832009,
        0.066594385,
        0.013972063,
        -0.018454744,
        -0.059638955,
        0.03856133,
        -0.01244498,
        -0.109063044,
        -0.00258201,
        0.04090507,
        0.07400475,
        0.026242465,
        -0.07876024,
        0.11164126,
        0.01099771,
        0.0131642185,
        -0.04036516,
        -0.009867459,
        -0.036595438,
        -0.093301944,
        0.02398446,
        -0.087460116,
        -0.07356979,
        -0.007095346,
        0.069862954,
        -0.027266523,
        -0.007665925,
        0.048004378,
        0.071703404,
        -0.0737683,
        -0.02775619,
        0.030077124,
        -0.024521783,
        -0.04143935,
        0.04218047,
        0.02220641,
        0.05151152,
        0.0024453597,
        -0.03300387,
        -0.04201622,
        -0.003383837,
        -0.043555696,
        -0.00885973,
        -0.09889512,
        0.045539405,
        0.048017535,
        -0.060423415,
        0.008561578,
        -0.06601728,
        0.087986246,
        -0.014303102,
        0.034236092,
        0.02674059,
        0.008994257,
        -0.0052561946,
        0.058496416,
        -0.06134325,
        -0.0959717,
        -0.6151156,
        0.04330559,
        0.017020028,
        -0.086301245,
        -0.15922125,
        -0.010756488,
        -0.038886514,
        0.07591362,
        0.07945725,
        -0.044241004,
        0.04467569,
        -0.049305163,
        -0.018688196,
        -0.046721306,
        -0.013694247,
        0.035915997,
        -0.03137749
      ]
    },
    {
      "id": "worldsfair-speaker-137",
      "conference": "worldsfair",
      "name": "Lance Martin",
      "role": "MTS",
      "company": "Anthropic",
      "twitter": "https://x.com/RLanceMartin",
      "talks": [
        {
          "title": "Claude for long-horizon tasks",
          "description": "Claude is capable of long horizon tasks. In this talk, we'll share lessons learned about building agent harnesses for reliable and secure long-horizon work. This include decoupling the brain and hands, self-verification, self-learning, and design for evolving agent harnesses.",
          "day": "Day 3 — Session Day 2",
          "time": "9:30am-9:50am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.049089458,
        0.06005781,
        -0.012938884,
        0.03472611,
        0.0745742,
        -0.04303793,
        -0.06042977,
        0.0062284977,
        0.071185805,
        -0.21521577,
        -0.012327103,
        -0.058393218,
        0.09690268,
        0.05276934,
        0.0072875596,
        0.03602834,
        0.10716349,
        -0.0726056,
        -0.08598556,
        -0.05738748,
        -0.033328597,
        -0.05756561,
        -0.033468563,
        0.07444876,
        -0.030209513,
        0.1429661,
        -0.060053725,
        0.07604689,
        -0.10237986,
        0.39647418,
        -0.091782086,
        0.0364876,
        -0.0023935158,
        -0.049242128,
        0.09458635,
        -0.07295204,
        -0.017467445,
        -0.011985029,
        0.04157632,
        -0.03761969,
        0.029928295,
        0.00616872,
        -0.009460804,
        0.03471096,
        -0.0032795845,
        0.022729857,
        -0.043776505,
        0.0010011848,
        -0.06822779,
        -0.032686707,
        -0.09664465,
        0.03363622,
        0.018184153,
        0.008166061,
        0.014651157,
        0.04956656,
        0.008745005,
        -0.036362443,
        0.07651587,
        0.0110984165,
        -0.054962642,
        -0.006169796,
        0.04676885,
        0.0055093216,
        0.012789998,
        -0.055026338,
        0.065366484,
        -0.044298436,
        -0.0003963598,
        -0.07589478,
        0.058807835,
        0.058795173,
        -0.09288406,
        -0.02236595,
        -0.050707903,
        -0.09756595,
        -0.06268182,
        0.06852389,
        -0.055810392,
        -0.054161303,
        0.024610851,
        0.06699383,
        -0.039646402,
        -0.08008637,
        0.09119569,
        -0.033579204,
        -0.07389781,
        0.05546245,
        -0.059204366,
        0.052539825,
        0.016042402,
        -0.02145937,
        -0.049748592,
        -0.0123518575,
        -0.062363226,
        -0.009630702,
        -0.07619584,
        0.062883,
        -0.038479745,
        0.026857525,
        -0.0012241,
        0.0038530033,
        0.04050557,
        -0.011163738,
        0.004331376,
        -0.0150520075,
        -0.031959288,
        0.016476551,
        0.09335088,
        -0.039844565,
        -0.09127019,
        -0.6444712,
        0.0110615,
        -0.029370416,
        -0.029186888,
        -0.16935341,
        0.02365606,
        -0.018474093,
        0.030668266,
        0.08151774,
        -0.032068126,
        0.042293575,
        -0.039726164,
        -0.06739852,
        -0.02957505,
        -0.017819997,
        0.091502704,
        0.0063727344
      ]
    },
    {
      "id": "worldsfair-speaker-138",
      "conference": "worldsfair",
      "name": "Antje Barth",
      "role": "Member of Technical Staff",
      "company": "Amazon AGI Lab",
      "twitter": "https://x.com/anbarth",
      "talks": [
        {
          "title": "Perception Agents",
          "description": "Human-to-agent collaboration is changing, becoming more visual. The agents most teams ship today still wait for us to type a paragraph to explain what we're looking at. They cannot see a screen, navigate a UI that changes, or recover when an application throws an unexpected modal. That is the architectural gap between agents that demo well and agents that work alongside real teams in real software. Perception agents close it. They see and use computers the way people do, reason about what they see, and act with clicks and keystrokes. We call this shared perception between humans and agents, and the perception agent harness is what makes it reliable in production.",
          "day": "Day 3 — Session Day 2",
          "time": "9:50am-10:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.03440417,
        0.022128992,
        0.03517605,
        0.004496668,
        0.026878402,
        -0.052479386,
        -0.09111269,
        -0.0072090537,
        0.044681538,
        -0.16654147,
        -0.048648298,
        -0.029481456,
        0.086617604,
        -0.0127602955,
        -0.066878915,
        -0.010536284,
        0.095311515,
        -0.09131557,
        -0.052500464,
        -0.09074276,
        0.0056920326,
        -0.03203491,
        -0.013441024,
        0.07369885,
        -0.0069261966,
        0.14921938,
        0.0078065637,
        0.009759089,
        -0.16125244,
        0.50383705,
        -0.080975726,
        0.008952511,
        -0.0064024194,
        -0.00672529,
        0.06387344,
        -0.073751576,
        0.042479005,
        -0.019902844,
        0.03093526,
        0.022076443,
        -0.0017937359,
        0.042049598,
        -0.07501252,
        0.026708309,
        -0.04737296,
        0.0024224904,
        -0.030298084,
        -0.011978745,
        0.014471507,
        0.0044257934,
        -0.031396028,
        -0.028740019,
        0.015738681,
        -0.060266998,
        0.057929397,
        0.025692811,
        0.06958764,
        -0.046148293,
        0.07297764,
        -0.010123833,
        -0.028010486,
        -0.020263245,
        -0.0053189998,
        0.031123253,
        0.01971688,
        -0.05267246,
        0.033281088,
        -0.062691905,
        -0.015039168,
        -0.06499955,
        -0.009428274,
        0.06143827,
        -0.089694954,
        0.012583821,
        -0.043545004,
        -0.1038639,
        -0.05107201,
        0.048662413,
        -0.014582309,
        -0.019395877,
        -0.034762044,
        0.042336103,
        -0.076027766,
        -0.025764475,
        0.03807451,
        -0.022934413,
        -0.10152565,
        0.011222231,
        -0.086924665,
        0.015658703,
        0.03499452,
        -0.02006881,
        -0.04468918,
        0.013062168,
        -0.08021693,
        0.042977236,
        -0.05124946,
        0.0701282,
        0.08791444,
        0.023350036,
        -0.032651745,
        -0.0492704,
        -0.0027031405,
        -0.010810393,
        0.012300514,
        -0.053274438,
        -0.01114835,
        0.029262634,
        0.07510744,
        -0.046436515,
        0.000555077,
        -0.60375863,
        -0.0023516354,
        0.0082646115,
        -0.0014657074,
        -0.18290442,
        0.035835244,
        -0.08287864,
        0.065910965,
        0.047061034,
        -0.015452697,
        0.08653954,
        -0.085949115,
        -0.019984419,
        -0.03641459,
        0.0075139175,
        0.023215387,
        0.025812611
      ]
    },
    {
      "id": "worldsfair-speaker-139",
      "conference": "worldsfair",
      "name": "Tariq Shaukat",
      "role": "Chief Executive Officer",
      "company": "Sonar",
      "twitter": "https://x.com/tariqshaukat",
      "talks": [
        {
          "title": "In the Land of AI Agents, the Verifiers Are King",
          "description": "As AI agents take on increasingly complex development tasks, the critical challenge has shifted from generation to verification. Hallucination is not a temporary bug. Evidence suggests that as models grow more capable, failures become more frequent and more convincing, making cognitive surrender among human reviewers…",
          "day": "Day 3 — Session Day 2",
          "time": "10:10am-10:30am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.02942185,
        0.028745925,
        0.012187687,
        0.032209765,
        0.038476516,
        -0.050369605,
        -0.101607345,
        0.015347598,
        0.048150267,
        -0.23082125,
        -0.040860876,
        -0.0000614595,
        -0.0032136734,
        -0.088738255,
        -0.053000335,
        0.014431632,
        0.038587857,
        -0.049673617,
        -0.074029006,
        -0.057682388,
        -0.00040992105,
        -0.05740593,
        0.04587552,
        0.07090313,
        -0.028537514,
        0.12221196,
        0.0039004667,
        -0.0036617678,
        -0.10455793,
        0.4716786,
        -0.017933715,
        0.0155534865,
        -0.0030603989,
        -0.039033502,
        0.0057054935,
        -0.0571497,
        -0.04078105,
        -0.05967099,
        0.035139658,
        0.04637752,
        -0.059725594,
        0.040306404,
        -0.046995915,
        0.081184514,
        0.06084826,
        -0.06845238,
        -0.051777106,
        -0.040907074,
        -0.012763325,
        -0.007761519,
        -0.008841268,
        -0.023354435,
        -0.024996135,
        -0.02688771,
        0.025303187,
        0.008628232,
        -0.020348012,
        -0.040574774,
        0.027915444,
        -0.018529939,
        -0.11999479,
        0.011117965,
        0.047121868,
        0.014525483,
        0.027530361,
        -0.12852843,
        0.020348962,
        -0.016341407,
        -0.0010584428,
        -0.060499463,
        -0.029790862,
        0.008640559,
        -0.08002326,
        0.058929108,
        -0.06417281,
        -0.11818433,
        -0.016759679,
        0.081046745,
        -0.047786843,
        -0.013173359,
        -0.007529947,
        0.034208395,
        -0.1632074,
        -0.0788516,
        0.056613583,
        0.01017463,
        -0.038108665,
        0.053821456,
        -0.009400626,
        0.08135794,
        0.024910329,
        0.0377313,
        -0.028945077,
        0.050110806,
        -0.110301,
        -0.02532009,
        -0.05949712,
        -0.015240687,
        -0.021509666,
        0.016931878,
        0.008052334,
        -0.0051516974,
        0.07161269,
        0.0075017284,
        0.06114365,
        0.010278373,
        0.05607837,
        -0.0446792,
        0.04020453,
        -0.012147884,
        -0.103387,
        -0.5922369,
        0.0692081,
        -0.011226633,
        -0.03982723,
        -0.15731788,
        0.017742235,
        -0.03553542,
        0.122624986,
        0.055604175,
        -0.0030447186,
        0.07533152,
        -0.078045554,
        0.00435685,
        -0.02443664,
        0.0681057,
        0.039362118,
        -0.013948968
      ]
    },
    {
      "id": "worldsfair-speaker-140",
      "conference": "worldsfair",
      "name": "Richard Socher",
      "role": "CEO & Co-Founder",
      "company": "You.com / Recursive Superintelligence",
      "twitter": "https://x.com/RichardSocher",
      "talks": [
        {
          "title": "First Steps Toward Automated AI Research",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.049523093,
        0.0114610605,
        0.021233335,
        0.13119188,
        0.05950823,
        -0.025830168,
        -0.06867442,
        0.013177475,
        0.08439117,
        -0.19017856,
        -0.07047068,
        -0.045876116,
        0.013513745,
        0.04662331,
        -0.057412796,
        0.08944542,
        0.030288717,
        -0.09386515,
        -0.02341872,
        -0.13866764,
        -0.036005404,
        -0.048433393,
        0.011314759,
        0.054198146,
        -0.038386032,
        0.051098794,
        0.016085112,
        0.034777522,
        -0.08808862,
        0.46005547,
        -0.08535437,
        -0.011850504,
        -0.04733962,
        -0.014636232,
        0.064001925,
        -0.010097233,
        -0.02017084,
        0.0026223073,
        0.082503706,
        0.09657258,
        -0.0049059587,
        0.020505939,
        -0.031174503,
        0.032344554,
        -0.014023031,
        -0.02322792,
        -0.031776536,
        -0.032637883,
        -0.036329634,
        -0.048851103,
        -0.07653288,
        0.07406968,
        -0.028513793,
        -0.055770326,
        0.036194943,
        0.09191211,
        -0.03669364,
        -0.100010864,
        0.06499126,
        -0.058843866,
        -0.16862951,
        -0.027815921,
        -0.0038545458,
        0.11797221,
        -0.01501298,
        -0.036593996,
        0.04095642,
        -0.06242811,
        0.021320058,
        -0.11596734,
        -0.052585293,
        0.0064971214,
        -0.029540155,
        0.01769206,
        -0.07286371,
        -0.0066183074,
        -0.017328018,
        0.05833158,
        -0.011102449,
        -0.029228,
        0.0038192163,
        0.027879538,
        -0.03914917,
        -0.066369765,
        -0.022627821,
        0.014908709,
        -0.003458002,
        0.04993542,
        -0.062029447,
        0.086780615,
        0.082956366,
        -0.020771047,
        -0.05379267,
        -0.017448641,
        -0.05341865,
        0.050617617,
        -0.07248818,
        0.05151166,
        0.106114745,
        -0.018729243,
        0.0035510233,
        -0.00034569865,
        0.04261387,
        -0.0492409,
        -0.024166325,
        -0.026481936,
        0.05760481,
        -0.035594303,
        0.032018453,
        0.009759275,
        -0.050382186,
        -0.6066526,
        0.0015931188,
        -0.003838571,
        0.0015604541,
        -0.16411628,
        -0.007897377,
        -0.031312305,
        0.06343468,
        0.051780492,
        -0.034695085,
        -0.029515086,
        0.019214666,
        0.042703997,
        0.02793087,
        0.06405663,
        0.031710595,
        0.011253816
      ]
    },
    {
      "id": "worldsfair-speaker-141",
      "conference": "worldsfair",
      "name": "Christopher Manning",
      "role": "Distinguished Member of Technical Staff at Moonlake AI; Thomas M. Siebel Professor in Machine Learning at Stanford University; General Partner at AIX Ventures",
      "company": "Moonlake AI",
      "twitter": "https://x.com/chrmanning",
      "talks": [
        {
          "title": "Building the simulation infrastructure for practical world model use",
          "description": "What is the most important capability for world model applications and the pursuit of embodied AI? We believe it is not a question of having the most beautiful pixels but the ability to reason about causality in multimodal environments. At Moonlake, we are working on building action-conditioned multimodal world models which provide spatial and physical state consistency over long time periods. We believe that building and training on synthetic worlds provides the data and compute efficient path to truly useful world models. We are building the simulation infrastructure platform for companies that need to build and manage worlds (assets, scenes, digital twins) at scale, including robotics/autonomy teams, digital factory operators, and game authors. Our product today primarily finds applicability in simulation and the operationalization of digital twins. Simulation can include training robotics, world models for AGI research, autonomous vehicles, or content creation for media and entertainment. Operationalization of digital twins involves the reconstruction of scans into reusable assets, e.g., turning image and point-cloud scans into sim ready assets for digital factory Integration projects. We are building toward a future where AI systems do not just generate worlds, but understand how they work. Moonlake learns from each workflow: The more workflows, failures, and human interventions that Moonlake sees, the better it becomes at reconstructing, validating, and preparing complex simulation worlds. The session will include discussion and demos.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "sponsor"
        },
        {
          "title": "Building the simulation infrastructure for practical world model use",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.059134282,
        0.02429401,
        -0.018303784,
        0.07444873,
        0.019637285,
        -0.04816467,
        -0.049519226,
        0.002896382,
        0.07173075,
        -0.25975272,
        -0.0112299835,
        -0.06918748,
        -0.04602839,
        -0.033774327,
        -0.08046336,
        0.029384065,
        0.052516956,
        -0.08163299,
        -0.06113254,
        -0.08703836,
        -0.05709636,
        -0.0562009,
        0.00046076666,
        0.14057486,
        -0.03977581,
        0.056985714,
        0.059054866,
        0.019800363,
        -0.05628602,
        0.42805764,
        -0.14538974,
        -0.020653786,
        -0.032175608,
        0.024260135,
        0.12672798,
        -0.040366095,
        -0.11867579,
        -0.0049201027,
        0.05397203,
        0.07070779,
        0.0071994616,
        0.030999826,
        -0.0056333914,
        0.027222212,
        -0.036669433,
        0.013883467,
        -0.016648611,
        0.0076090177,
        -0.0055044605,
        0.03404944,
        -0.056186203,
        -0.010700012,
        0.02415496,
        0.0048678536,
        0.044797935,
        -0.01889898,
        0.08590641,
        -0.01862355,
        0.09556899,
        -0.030393938,
        -0.1405466,
        0.012040855,
        0.047494996,
        0.026348135,
        -0.008913857,
        -0.018911468,
        0.014894166,
        -0.031331003,
        0.01197313,
        -0.062024042,
        0.03860172,
        0.027226562,
        -0.078523956,
        0.043369394,
        -0.055213794,
        -0.08595435,
        -0.06826724,
        0.046569444,
        -0.067227565,
        -0.017698318,
        -0.005105027,
        0.105003275,
        0.015256657,
        -0.05423914,
        0.00488177,
        -0.027293527,
        -0.05359537,
        0.04486268,
        -0.06917123,
        0.0689368,
        -0.024944909,
        -0.058900032,
        -0.061377328,
        -0.02802263,
        -0.041130964,
        0.034302693,
        -0.10081361,
        0.04656878,
        0.033732414,
        0.0112138605,
        0.033183847,
        0.027778342,
        0.12477893,
        -0.06966749,
        0.017802665,
        -0.015240874,
        0.031937014,
        0.026845051,
        0.05074419,
        -0.054638334,
        -0.06525895,
        -0.6093168,
        0.022537233,
        -0.0014503613,
        -0.037846755,
        -0.13042265,
        0.00916018,
        -0.084856324,
        0.053975653,
        -0.0037982021,
        -0.0030293616,
        0.008009163,
        0.020675896,
        -0.069003955,
        -0.03431165,
        0.0016246123,
        0.07525445,
        0.024906825
      ]
    },
    {
      "id": "worldsfair-speaker-142",
      "conference": "worldsfair",
      "name": "Parth Asawa",
      "role": "CS PhD student",
      "company": "UC Berkeley",
      "twitter": "https://x.com/pgasawa",
      "talks": [
        {
          "title": "Continual Learning Bench",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.034550462,
        0.034111883,
        0.027581647,
        0.034940943,
        0.025531933,
        -0.044035867,
        -0.0796854,
        -0.020828733,
        0.027635992,
        -0.19156818,
        0.00023632236,
        -0.019413453,
        0.06188369,
        0.05081304,
        -0.004610015,
        -0.06374683,
        0.04076976,
        -0.07187293,
        -0.043314908,
        -0.06286747,
        -0.07021022,
        -0.075165905,
        -0.075208895,
        0.024348686,
        -0.09184229,
        0.044920873,
        0.027532281,
        0.047757577,
        -0.11493162,
        0.45554966,
        -0.0720921,
        -0.100189164,
        0.011338336,
        0.046357393,
        0.02738348,
        -0.019321103,
        0.06041606,
        0.03229563,
        0.022163259,
        0.06883209,
        -0.031411882,
        0.019531697,
        -0.039073773,
        0.07386896,
        -0.04355754,
        0.00075563474,
        -0.10673556,
        -0.04760539,
        -0.032077383,
        -0.025516355,
        -0.06914471,
        -0.0143613685,
        0.042643428,
        -0.011498871,
        0.0042306404,
        0.06293193,
        -0.020263238,
        -0.08881838,
        0.07259422,
        0.0033221992,
        -0.11140385,
        0.0029811456,
        0.024796199,
        -0.008815535,
        -0.014939733,
        -0.0052995197,
        0.07749287,
        -0.032579355,
        -0.051538453,
        0.008958446,
        0.10476936,
        0.021606581,
        -0.073868565,
        0.016012175,
        -0.07167518,
        -0.032630675,
        -0.032979205,
        0.13797317,
        -0.018355433,
        0.070634745,
        0.0045913397,
        0.06972211,
        -0.019568717,
        -0.05988052,
        0.07671589,
        -0.008440531,
        0.01727771,
        0.05348073,
        -0.004108558,
        0.044222906,
        0.014730757,
        -0.062506795,
        -0.04464325,
        -0.033227485,
        -0.019783994,
        0.07104285,
        -0.06943552,
        0.07081831,
        -0.018224377,
        0.03585354,
        -0.010811996,
        -0.08732809,
        0.0149765005,
        0.01926982,
        -0.03278552,
        -0.06939257,
        0.03420077,
        0.010566907,
        0.011419856,
        -0.024726002,
        -0.061921183,
        -0.62905675,
        -0.026246317,
        0.035984278,
        -0.030982235,
        -0.17023313,
        0.01899781,
        -0.039065022,
        0.011771286,
        0.11799161,
        -0.07481771,
        0.039283514,
        0.0064641126,
        0.03965219,
        -0.030041544,
        -0.040051512,
        0.06462577,
        -0.021478176
      ]
    },
    {
      "id": "worldsfair-speaker-143",
      "conference": "worldsfair",
      "name": "Thor 雷神 Schaeff",
      "role": "Developer Experience Engineer",
      "company": "Google DeepMind",
      "twitter": "https://x.com/thorwebdev",
      "talks": [
        {
          "title": "Build realtime multimodal agents with Gemini Live",
          "description": "The Gemini Live API is incredible versatile when it comes to building realtime AI experiences. From live translation across 2000 different language pairs to building realtime multimodal agents that can work across text, audio, and vision. This workshop gets you from zero to fully conversational agent in a matter of hours.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "Build realtime multimodal agents with Gemini Live (continued 2)",
          "description": "The Gemini Live API is incredible versatile when it comes to building realtime AI experiences. From live translation across 2000 different language pairs to building realtime multimodal agents that can work across text, audio, and vision. This workshop gets you from zero to fully conversational agent in a matter of hours.",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        },
        {
          "title": "Build realtime multimodal agents with Gemini Live (continued 3)",
          "description": "The Gemini Live API is incredible versatile when it comes to building realtime AI experiences. From live translation across 2000 different language pairs to building realtime multimodal agents that can work across text, audio, and vision. This workshop gets you from zero to fully conversational agent in a matter of hours.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        },
        {
          "title": "Build realtime multimodal agents with Gemini Live (continued 4)",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04712621,
        0.08662442,
        0.039763678,
        0.030118806,
        0.05390273,
        -0.048125774,
        -0.04693898,
        -0.017547382,
        0.06844608,
        -0.26749393,
        -0.054740723,
        0.04966429,
        0.0041257935,
        -0.023848578,
        -0.080981396,
        0.067033775,
        0.124821596,
        -0.13579093,
        -0.08080614,
        -0.05390196,
        -0.012745423,
        -0.0931208,
        -0.06290098,
        0.035760615,
        -0.026352948,
        0.108452976,
        0.005633126,
        0.0032479716,
        -0.08403831,
        0.44876418,
        -0.111935265,
        -0.045631997,
        0.029643938,
        -0.025685504,
        0.08695541,
        -0.015281641,
        0.0020842694,
        -0.035143066,
        0.05441741,
        0.025294233,
        -0.037577193,
        0.048089128,
        0.010508398,
        -0.010746366,
        -0.034954067,
        -0.036948606,
        -0.0629909,
        0.026414901,
        0.089727394,
        0.014135293,
        -0.044285323,
        0.009725134,
        -0.009928423,
        -0.046270575,
        0.03260127,
        0.016293308,
        0.049177278,
        -0.058202405,
        0.111979194,
        0.0073758676,
        -0.07967752,
        -0.053582244,
        0.015353402,
        -0.0001469941,
        0.012780549,
        -0.0363021,
        0.021644993,
        -0.039607216,
        -0.046077468,
        -0.08086591,
        0.038748506,
        -0.0042744824,
        -0.03063237,
        0.048107684,
        -0.036330353,
        -0.1029663,
        -0.06689581,
        0.05407545,
        -0.04279195,
        -0.021681922,
        -0.029734703,
        0.041877702,
        -0.13297047,
        -0.025630446,
        0.03212395,
        0.041170992,
        -0.08610539,
        0.003454133,
        -0.012979359,
        0.05842666,
        -0.037110604,
        0.062060878,
        -0.1259749,
        0.07779675,
        -0.058736276,
        -0.03369812,
        -0.015368043,
        0.014156253,
        0.021494836,
        -0.029241096,
        0.023758842,
        -0.041664794,
        0.009007842,
        -0.022827512,
        -0.00038132942,
        -0.055837154,
        -0.030760963,
        0.01945612,
        0.047509074,
        -0.058544356,
        -0.038871832,
        -0.5860416,
        0.032834988,
        -0.015884504,
        -0.072886884,
        -0.15850523,
        -0.0082649365,
        -0.060729563,
        -0.0021756766,
        0.07706618,
        -0.055871855,
        0.08274701,
        0.034696978,
        -0.045301024,
        -0.04253604,
        0.071685985,
        -0.0098489,
        -0.0049058385
      ]
    },
    {
      "id": "worldsfair-speaker-144",
      "conference": "worldsfair",
      "name": "Andon Labs",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Vending-Bench: Long-Horizon Agent Evals for a Simulated Vending Business",
          "description": "Long-horizon agent evals via a simulated vending machine business, testing negotiation, pricing, and supplier management over 365 days.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.04883238,
        0.07210425,
        0.044788323,
        0.009236126,
        0.0171622,
        0.011640642,
        -0.067684196,
        -0.0050043724,
        0.0070102047,
        -0.2771027,
        -0.0111102,
        -0.043725662,
        0.058660466,
        0.01948022,
        -0.025557509,
        0.05730898,
        0.049524713,
        -0.102912664,
        -0.044511802,
        -0.09414583,
        -0.10466674,
        -0.03569426,
        -0.03807802,
        -0.020283114,
        -0.03357576,
        0.100580886,
        -0.04049602,
        0.034874693,
        -0.040227685,
        0.43410918,
        -0.081872545,
        0.04147562,
        -0.036471006,
        -0.017452454,
        0.021263773,
        -0.068998724,
        -0.009485079,
        -0.009895985,
        0.05424891,
        -0.0014799546,
        0.02712986,
        -0.030261727,
        -0.049077835,
        0.053767927,
        -0.07375653,
        -0.020691952,
        -0.019634591,
        -0.008910285,
        -0.021296894,
        0.053769145,
        -0.08533226,
        -0.022601856,
        0.06978536,
        -0.00247663,
        -0.0023224934,
        0.023336394,
        0.054626696,
        -0.051065177,
        0.11521552,
        0.03478357,
        -0.10244678,
        0.0033456422,
        -0.02589841,
        0.0006758743,
        -0.08715032,
        -0.0031594404,
        0.10596449,
        -0.09228926,
        0.014494526,
        -0.07952465,
        0.04993357,
        0.05453106,
        -0.036264293,
        0.05407023,
        -0.0754178,
        -0.11139559,
        -0.074528776,
        0.06604929,
        0.00054649654,
        0.05565669,
        0.03014668,
        0.117530055,
        0.00013440124,
        -0.06350721,
        0.021199679,
        0.029520318,
        -0.019364614,
        0.038396847,
        -0.04504925,
        0.083978176,
        0.011129891,
        0.004281592,
        -0.053668384,
        -0.050160784,
        -0.014576169,
        0.060817085,
        -0.08161307,
        0.030761963,
        -0.05939002,
        0.023568869,
        0.025868222,
        -0.020118525,
        0.089618854,
        0.0065067187,
        0.00052038115,
        0.057881925,
        0.011514414,
        0.037365034,
        0.024105964,
        -0.037066128,
        -0.056669388,
        -0.61055565,
        -0.07634419,
        -0.06326817,
        -0.022965558,
        -0.1578902,
        0.05305535,
        0.04472233,
        0.02349087,
        0.043093443,
        -0.03550462,
        0.07967523,
        -0.0005232475,
        -0.035239354,
        0.019161135,
        -0.03536137,
        0.07208135,
        0.0030855103
      ]
    },
    {
      "id": "worldsfair-speaker-145",
      "conference": "worldsfair",
      "name": "Geoffrey Litt",
      "role": "Design Engineer",
      "company": "Notion",
      "twitter": "https://x.com/geoffreylitt",
      "talks": [
        {
          "title": "On Malleable Software",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.014153484,
        0.041936375,
        0.011200289,
        -0.0027168742,
        0.07245422,
        -0.08342448,
        -0.0837981,
        0.02684016,
        0.10007645,
        -0.20004794,
        -0.05082834,
        -0.03515757,
        0.029143589,
        -0.016160186,
        0.015068986,
        0.008357687,
        0.009286352,
        -0.086591676,
        -0.018924879,
        -0.07154163,
        -0.052048165,
        -0.09013952,
        -0.0690168,
        0.040087085,
        -0.06966049,
        0.10969538,
        0.04196652,
        0.027364299,
        -0.15144065,
        0.43133146,
        -0.06905723,
        0.028664175,
        0.07132393,
        0.023844581,
        0.10072855,
        -0.047881607,
        -0.021532092,
        0.001824697,
        0.07389531,
        0.030121477,
        -0.0035292099,
        0.047636878,
        -0.007053259,
        -0.012160599,
        -0.038904,
        -0.008813538,
        -0.093905576,
        -0.018896151,
        0.014301154,
        -0.017507382,
        -0.09956377,
        0.05319903,
        0.0077185603,
        -0.05352651,
        -0.006834944,
        -0.011431145,
        -0.049909268,
        -0.05972974,
        0.08976592,
        -0.02892528,
        -0.10852088,
        -0.050228074,
        0.005446264,
        0.035488423,
        0.0076786187,
        -0.038319543,
        0.10000085,
        0.002425705,
        -0.01252585,
        -0.07936062,
        0.03720196,
        -0.010371783,
        0.009756926,
        0.032269742,
        -0.034513492,
        -0.020831112,
        -0.02071771,
        0.050109774,
        -0.024878694,
        0.046014734,
        0.07640874,
        0.038365647,
        -0.099169485,
        -0.05718062,
        0.014982869,
        -0.08365385,
        -0.05122805,
        0.11413802,
        0.012825783,
        0.11339323,
        0.080206074,
        -0.029832557,
        -0.024422042,
        0.05028765,
        -0.043993674,
        0.058283757,
        -0.054927733,
        0.10569854,
        -0.024737922,
        0.032654386,
        -0.08747684,
        -0.020155592,
        0.06155421,
        -0.00674666,
        -0.060502097,
        -0.08832839,
        -0.0215001,
        0.013675349,
        0.034301616,
        -0.00095465826,
        0.04022075,
        -0.6250908,
        -0.020709764,
        -0.0065836594,
        -0.0014510723,
        -0.14838512,
        0.020413693,
        -0.017088097,
        0.050837316,
        0.06296888,
        0.011276477,
        0.0038969673,
        -0.025366062,
        -0.036425773,
        0.012620985,
        0.03805818,
        0.07752845,
        -0.022172943
      ]
    },
    {
      "id": "worldsfair-speaker-146",
      "conference": "worldsfair",
      "name": "Pierluca D'Oro",
      "role": "Founder",
      "company": "Stealth (formerly Meta)",
      "twitter": "https://x.com/proceduralia",
      "talks": [
        {
          "title": "Computer Use at the Edge of the Statistical Precipice",
          "description": "Evaluating Computer Use Agents (CUAs) on interactive environments is fraught with methodological pitfalls that the field has yet to systematically address. We show that a 1MB replay script that blindly executes a recorded action sequence without ever observing the screen outperforms frontier models on prominent static benchmarks, and prove that its expected success rate is exactly equal to the source agent's pass@k in deterministic environments. We trace this and other failures to two root causes: non-principled environment design (static, unsandboxed, or unreliably verified environments) and non-principled evaluation methodology (naive aggregation and misuse of pass@k for stateful UI interactions). To address the first, we propose PRISM, five design principles for CUA environments and instantiate them in DigiWorld, a benchmark of 15 realistic sandboxed mobile applications able to evaluate agents in over 3.2 million verified unique configurations. To address the second, we develop an aggregation framework that correctly accounts for the nested structure of CUA benchmarks. All together, we show that principled environment design and rigorous evaluation methodology are not optional refinements but prerequisites for meaningful CUA research.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.05435449,
        0.080143504,
        0.060317013,
        0.02884051,
        -0.063034385,
        -0.030515485,
        -0.048217498,
        0.029457038,
        0.004425373,
        -0.25461167,
        -0.039399367,
        -0.038801815,
        0.033252932,
        -0.0028094107,
        -0.07972869,
        -0.018806113,
        0.13256377,
        0.0029557182,
        -0.10945144,
        -0.03752582,
        -0.038163885,
        -0.11111068,
        -0.014953771,
        0.07196361,
        -0.061550573,
        0.08385134,
        0.097734086,
        0.079243526,
        -0.09916711,
        0.48560503,
        -0.15019366,
        -0.015322256,
        0.049685698,
        -0.03702675,
        0.059843797,
        -0.05958388,
        -0.003046136,
        0.030218627,
        0.06488949,
        0.036990866,
        0.060824245,
        0.036955323,
        -0.06767041,
        -0.009206974,
        -0.01463272,
        -0.0480831,
        -0.04160698,
        -0.0030723242,
        0.0079105925,
        -0.05789362,
        -0.08089985,
        0.031442117,
        0.059078086,
        0.014762155,
        0.03372632,
        0.025101392,
        -0.0017783139,
        -0.05891502,
        0.11333642,
        -0.008736376,
        -0.0986293,
        0.0034196118,
        0.0029102536,
        0.034728333,
        -0.0022839876,
        -0.04550573,
        0.02114859,
        -0.04219658,
        0.029521914,
        -0.032649163,
        0.026182566,
        0.016704721,
        -0.06921205,
        -0.047552254,
        -0.03647045,
        -0.06176287,
        -0.026575264,
        0.011071233,
        -0.01259462,
        0.019653054,
        -0.00016365272,
        0.0681687,
        0.0044810525,
        -0.05066131,
        0.012280346,
        -0.00964663,
        -0.012534329,
        0.035516027,
        -0.095240034,
        0.04096897,
        0.022698617,
        -0.12287143,
        -0.08372428,
        -0.014961114,
        -0.072306305,
        -0.050336,
        -0.050617073,
        0.064398326,
        -0.008355256,
        0.024017207,
        -0.021797951,
        -0.04662928,
        0.015527866,
        -0.021682825,
        0.017377766,
        -0.048747208,
        0.014056547,
        0.012957248,
        0.06635081,
        -0.07670778,
        -0.022531522,
        -0.57964385,
        -0.0122141065,
        0.026880808,
        -0.02281064,
        -0.15793575,
        0.030731305,
        -0.03319157,
        -0.005940745,
        0.049753677,
        -0.07815933,
        0.060108412,
        0.0015234004,
        0.07014116,
        -0.039806,
        0.0374546,
        0.038106706,
        -0.016554888
      ]
    },
    {
      "id": "worldsfair-speaker-147",
      "conference": "worldsfair",
      "name": "Kurtis Van Gent",
      "role": "Senior Staff Software Engineer",
      "company": "Google",
      "twitter": "",
      "talks": [
        {
          "title": "Build-Time vs. Run-Time: Why Your Dev Tools Will Fail in Production",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.053070985,
        0.052577037,
        0.009357371,
        0.0480843,
        0.04234786,
        -0.04031936,
        -0.08984503,
        -0.028064203,
        0.07445498,
        -0.18893176,
        -0.02472639,
        0.00060761947,
        0.028117266,
        -0.022971747,
        -0.025589919,
        0.04733636,
        0.12686992,
        -0.0799902,
        -0.0689116,
        -0.14054693,
        -0.02178285,
        -0.10884264,
        -0.016393302,
        0.036982775,
        0.011426824,
        0.08469804,
        -0.040702034,
        0.045620304,
        -0.07375642,
        0.5020285,
        -0.068039276,
        0.0348388,
        0.022466933,
        -0.03650987,
        0.03548702,
        -0.041770004,
        -0.047214337,
        0.05148683,
        -0.045880236,
        0.030601298,
        -0.06359125,
        -0.0029691495,
        -0.057486944,
        -0.021169221,
        -0.09681465,
        0.005231624,
        -0.09166058,
        -0.072606154,
        -0.051566914,
        -0.013668491,
        -0.076111816,
        0.05220515,
        -0.013746569,
        -0.081497,
        -0.018160388,
        -0.041668717,
        -0.07540387,
        -0.030076217,
        0.041978605,
        0.0320542,
        -0.09275484,
        0.024395924,
        0.0014410145,
        0.01119581,
        0.022107184,
        -0.07550987,
        0.044043295,
        -0.033270102,
        -0.04680504,
        0.04526062,
        0.010907155,
        0.0063876226,
        -0.04543417,
        0.03154558,
        -0.06774663,
        -0.015125379,
        -0.033459198,
        0.086712815,
        -0.064792044,
        -0.023377988,
        0.08397869,
        0.051416688,
        -0.07205394,
        -0.010658238,
        0.07276295,
        -0.015447445,
        -0.10144839,
        -0.011512134,
        0.00322007,
        0.02020415,
        0.052270982,
        -0.026632423,
        -0.07190696,
        -0.052282248,
        0.018862315,
        0.029563077,
        -0.062813185,
        0.020950513,
        -0.008582914,
        -0.0037666587,
        -0.03820484,
        -0.058466505,
        -0.019122005,
        -0.051891662,
        0.0030126465,
        -0.054973476,
        -0.081087895,
        -0.009289947,
        -0.01003946,
        -0.054437697,
        -0.049140573,
        -0.5866497,
        -0.014720646,
        -0.07676112,
        0.0210716,
        -0.18537165,
        -0.008968413,
        -0.07239176,
        0.022574093,
        0.09103211,
        -0.011194959,
        0.007921996,
        -0.005455307,
        -0.017747996,
        -0.02016814,
        0.090713784,
        0.030085498,
        -0.0076996414
      ]
    },
    {
      "id": "worldsfair-speaker-148",
      "conference": "worldsfair",
      "name": "Prerna Kakkar",
      "role": "",
      "company": "Google",
      "twitter": "",
      "talks": [
        {
          "title": "Build-Time vs. Run-Time: Why Your Dev Tools Will Fail in Production",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.0627333,
        0.069308355,
        0.01817373,
        0.0024444293,
        0.056061056,
        -0.06939894,
        -0.11340364,
        0.0012952668,
        0.021023681,
        -0.15919486,
        -0.029370368,
        0.042425063,
        0.03763199,
        -0.041087158,
        -0.06695012,
        0.067903966,
        0.09057257,
        -0.02999086,
        -0.058165837,
        -0.12042237,
        -0.028326707,
        -0.116265684,
        0.032341126,
        0.035635915,
        -0.015441659,
        0.07205361,
        -0.032328684,
        -0.050033353,
        -0.097610824,
        0.4964219,
        -0.10535491,
        0.061058298,
        -0.008395828,
        -0.07660965,
        0.0516759,
        -0.04035849,
        -0.0073589906,
        0.055158664,
        -0.024155857,
        0.04459309,
        -0.082549125,
        0.025937704,
        -0.0769892,
        -0.011394682,
        -0.01111783,
        -0.06260192,
        -0.026188422,
        -0.03880848,
        0.029859975,
        -0.014708446,
        -0.020381853,
        0.0019369181,
        0.006192414,
        -0.10145864,
        0.019517418,
        0.010222492,
        -0.08926951,
        -0.09035052,
        0.071183376,
        -0.01758471,
        -0.10022779,
        0.013239335,
        0.0065868795,
        0.015611734,
        -0.0034074606,
        -0.05004426,
        0.06412717,
        -0.07949627,
        -0.003170185,
        0.023787934,
        0.048931185,
        -0.0209104,
        -0.03455608,
        0.0958406,
        -0.052372456,
        -0.003970575,
        -0.021922734,
        0.13510711,
        -0.007559959,
        -0.007008649,
        0.06487858,
        0.0025316863,
        -0.087754786,
        -0.03112096,
        0.022617416,
        -0.0064777173,
        -0.048752084,
        0.037777234,
        -0.06386947,
        0.047061156,
        0.07238256,
        0.00053932803,
        -0.058809184,
        0.034075614,
        -0.019621318,
        0.031038657,
        -0.06471001,
        0.09698012,
        -0.009053135,
        0.06892042,
        -0.024886465,
        -0.040995747,
        -0.045210652,
        -0.030411359,
        -0.059291176,
        -0.018774351,
        -0.11080833,
        -0.01994511,
        0.0016269808,
        -0.07825237,
        -0.042155772,
        -0.5659514,
        0.101732686,
        -0.0153053375,
        0.00050305773,
        -0.15795948,
        0.0012151953,
        -0.06575478,
        -0.017330434,
        0.14981464,
        0.004197459,
        0.07459561,
        -0.030720633,
        0.03576173,
        -0.008394432,
        0.055431932,
        0.023010265,
        0.020378387
      ]
    },
    {
      "id": "worldsfair-speaker-149",
      "conference": "worldsfair",
      "name": "Diogo Almeida",
      "role": "Co-founder and CEO",
      "company": "TypeSafe AI",
      "twitter": "https://x.com/CompleteSkeptic",
      "talks": [
        {
          "title": "What's next after RLHF?",
          "description": "RLHF was a massive commercial success: roughly 100% of LLM usage is through RLHF’d models - but it was in many ways also a research failure. Let’s talk about how it conquered the world, how it defied its creators expectations, why AI is in the bimodal state it’s in (is it a bubble or a machine god?), and how to make AI actually transform the economy.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0012214434,
        0.0020368395,
        -0.02593161,
        0.03461992,
        0.04779871,
        0.004265699,
        -0.05838323,
        -0.024927212,
        0.078850046,
        -0.23329431,
        0.017565684,
        -0.052160095,
        0.03806127,
        -0.021556133,
        -0.068368606,
        0.028145546,
        0.0651437,
        -0.08913358,
        -0.011556844,
        -0.07444677,
        -0.08330702,
        -0.041794885,
        -0.0375241,
        0.1166871,
        -0.029985808,
        0.07720201,
        0.076917656,
        0.09311955,
        -0.05877649,
        0.40782732,
        -0.10583978,
        -0.075466506,
        0.0018224267,
        -0.02565397,
        0.0722612,
        -0.06256711,
        0.016121581,
        -0.020531006,
        0.013203386,
        0.113362335,
        0.015948987,
        0.09851078,
        -0.010585292,
        0.063434914,
        -0.01751871,
        -0.055157118,
        -0.045926493,
        0.008575345,
        -0.015075031,
        -0.007371709,
        -0.081344016,
        0.012645511,
        0.0042161597,
        -0.0019097099,
        0.014348878,
        -0.0068777585,
        0.025978992,
        -0.12152128,
        0.12807824,
        0.035553925,
        -0.14918514,
        0.021848373,
        0.02383416,
        -0.01951211,
        0.02817343,
        -0.054308172,
        -0.008424132,
        -0.10844213,
        0.045593567,
        -0.03235254,
        0.028462106,
        0.010854454,
        -0.039107498,
        -0.021877471,
        -0.10802089,
        -0.12064155,
        -0.03236894,
        0.037185524,
        0.039013244,
        -0.018348424,
        0.022706576,
        -0.027704004,
        -0.07582241,
        -0.023815488,
        0.048575617,
        -0.03991211,
        -0.022954952,
        0.038110595,
        0.01727283,
        0.07983758,
        0.074059136,
        -0.05761712,
        -0.08508214,
        0.042867176,
        -0.0723577,
        -0.0004260863,
        -0.049378064,
        0.046392825,
        -0.00068730273,
        -0.033566054,
        -0.054951403,
        -0.022786286,
        0.03925517,
        0.001868722,
        0.010909745,
        -0.044567056,
        0.090653844,
        0.027850077,
        0.09766803,
        -0.02564785,
        -0.061011165,
        -0.61601204,
        -0.037427694,
        -0.0054053883,
        0.008538405,
        -0.18271276,
        -0.0070567266,
        -0.011057156,
        -0.00020444518,
        0.061521254,
        -0.01963659,
        0.05539567,
        0.00992754,
        0.031237755,
        -0.024405675,
        0.08588481,
        -0.003298144,
        -0.04087492
      ]
    },
    {
      "id": "worldsfair-speaker-150",
      "conference": "worldsfair",
      "name": "Varun Shenoy",
      "role": "Cofounder",
      "company": "Long Lake",
      "twitter": "https://x.com/varunshenoy_",
      "talks": [
        {
          "title": "Vertical Superintelligence: Making AI Work in America's Messiest Industries",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.004565246,
        0.07085047,
        0.010394703,
        0.07513131,
        -0.008472039,
        -0.0688749,
        -0.10503967,
        0.102685705,
        0.07078213,
        -0.18011895,
        -0.0041502793,
        -0.1269256,
        0.015496125,
        -0.059043594,
        -0.10478352,
        -0.0066275457,
        -0.0045168702,
        -0.10413873,
        -0.044031806,
        -0.041711748,
        0.019267896,
        -0.12589084,
        -0.05176502,
        0.06656822,
        -0.06650146,
        0.08031505,
        -0.032604933,
        0.023376307,
        -0.06209414,
        0.4565643,
        0.018339772,
        0.000799945,
        -0.0060120705,
        0.010827021,
        0.04068705,
        -0.00800081,
        0.014513469,
        0.045883384,
        -0.022492923,
        0.097667255,
        -0.009780065,
        -0.00035437258,
        -0.01920137,
        0.005553227,
        0.04616473,
        -0.026568294,
        -0.036572803,
        0.0038369254,
        0.011632279,
        -0.018214528,
        -0.06812719,
        -0.049473163,
        0.0674392,
        -0.0510755,
        -0.011194118,
        0.07110405,
        0.027895072,
        -0.012300042,
        0.028623614,
        -0.082717486,
        -0.14803328,
        0.020991223,
        0.021653615,
        0.06786223,
        -0.0023260107,
        -0.07531621,
        0.015025602,
        -0.06055948,
        -0.017578157,
        0.018190611,
        0.059994258,
        -0.036789577,
        -0.010375659,
        0.014855429,
        -0.056455236,
        -0.063551016,
        0.005364854,
        0.08029675,
        -0.0021477467,
        -0.032745007,
        0.06880985,
        0.07794688,
        -0.07560552,
        -0.06743613,
        -0.039754555,
        -0.091617,
        -0.058095716,
        0.09546504,
        -0.056807607,
        0.1084003,
        0.009844221,
        0.0057498408,
        -0.08119859,
        0.011022253,
        0.005010662,
        -0.029036928,
        -0.10350716,
        0.06809595,
        -0.05498464,
        -0.017541854,
        0.00085200026,
        0.012040486,
        0.0139646605,
        -0.022878107,
        -0.06365235,
        -0.029692698,
        0.029982716,
        -0.03646228,
        0.042808972,
        0.017712338,
        -0.08027437,
        -0.6123684,
        -0.009169287,
        0.03078162,
        0.017518045,
        -0.15728168,
        0.04666961,
        -0.039200146,
        0.114260554,
        0.04131131,
        -0.014146131,
        0.041204654,
        0.006440996,
        0.015744831,
        0.023053331,
        0.0035354693,
        0.04881769,
        0.01384521
      ]
    },
    {
      "id": "worldsfair-speaker-151",
      "conference": "worldsfair",
      "name": "Rasmus Wissmann",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Vertical Superintelligence: Making AI Work in America's Messiest Industries",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.022257177,
        0.029329386,
        0.023178106,
        0.022401506,
        0.023617938,
        -0.08936263,
        -0.021566322,
        0.011169851,
        0.045059677,
        -0.2276834,
        -0.07195851,
        -0.0008364792,
        0.006811743,
        -0.014282633,
        -0.11809534,
        -0.024873225,
        0.090827085,
        -0.037340604,
        -0.005123766,
        -0.077119134,
        -0.017382475,
        -0.021475198,
        -0.07263161,
        0.052141067,
        -0.038750567,
        0.08776103,
        -0.054449376,
        0.027460843,
        -0.051629577,
        0.44935822,
        -0.04155185,
        0.025741693,
        -0.0020869356,
        0.05937341,
        0.054745823,
        -0.06436581,
        -0.038512938,
        -0.04345561,
        0.053655434,
        0.13625036,
        -0.032520603,
        0.06985223,
        -0.01618393,
        0.0062563093,
        0.04340255,
        -0.021996295,
        -0.061329536,
        -0.060556117,
        0.027710028,
        -0.061346937,
        -0.10968102,
        -0.021087859,
        0.018499527,
        -0.03026544,
        0.051291097,
        0.039464533,
        -0.015966248,
        -0.039634123,
        0.10233349,
        -0.05853335,
        -0.08731401,
        -0.017329572,
        -0.002827494,
        0.0646724,
        0.0809986,
        -0.043891024,
        0.049563557,
        -0.0636796,
        -0.010515093,
        -0.026240483,
        -0.019865362,
        -0.026831407,
        -0.11503853,
        -0.015996387,
        -0.07740777,
        -0.03200336,
        -0.04130185,
        0.023770535,
        -0.0019927807,
        -0.030206513,
        0.036721863,
        0.034933418,
        -0.04967006,
        -0.04661715,
        0.037281647,
        -0.014013333,
        -0.07690977,
        0.05740167,
        0.0119274715,
        0.05605062,
        0.106668875,
        -0.062029973,
        -0.05617807,
        0.0064397044,
        -0.03218366,
        -0.013133566,
        -0.0996852,
        0.02783571,
        0.038462166,
        0.011253122,
        0.0068380157,
        0.0033765866,
        0.01721144,
        0.03912919,
        0.019763915,
        -0.049337704,
        -0.0059488704,
        -0.035712816,
        0.0569689,
        0.041825015,
        -0.039388277,
        -0.6278419,
        -0.038780317,
        0.077757224,
        -0.021697335,
        -0.19155326,
        0.0024572616,
        0.014779168,
        0.07316266,
        0.023411402,
        -0.06574105,
        0.016957998,
        -0.0032762587,
        -0.044918716,
        0.019369606,
        0.07117596,
        0.0627674,
        0.032682877
      ]
    },
    {
      "id": "worldsfair-speaker-152",
      "conference": "worldsfair",
      "name": "Alex Volkov",
      "role": "",
      "company": "W&B from CoreWeave",
      "twitter": "https://x.com/altryne",
      "talks": [
        {
          "title": "The Z/L Continuum: Should AI Engineers Still Read Code?",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.06798879,
        -0.0005668871,
        0.09571118,
        0.028945314,
        -0.05455837,
        0.034675322,
        -0.024818165,
        -0.022790907,
        0.05196223,
        -0.23475707,
        -0.06469159,
        -0.038781267,
        0.026829077,
        -0.019436913,
        -0.010455955,
        0.001808772,
        0.107738696,
        -0.0997499,
        -0.015527567,
        -0.07879501,
        -0.07827297,
        -0.09204345,
        -0.023133943,
        -0.018675646,
        -0.043354046,
        0.121569864,
        0.0024884185,
        0.06669979,
        -0.11032538,
        0.49033183,
        -0.08038087,
        -0.014405475,
        0.042252287,
        -0.036403283,
        0.046385575,
        -0.10195586,
        -0.04808023,
        0.047944564,
        -0.012988334,
        0.08824227,
        -0.025783481,
        -0.0057644,
        -0.049117655,
        -0.0011393166,
        -0.07185554,
        -0.0049650054,
        -0.07324272,
        -0.025589725,
        0.0065026623,
        -0.07165026,
        -0.04594507,
        0.029983925,
        0.086487785,
        0.00014111312,
        0.0072067785,
        -0.0028370116,
        0.0020291829,
        -0.07835146,
        0.038862467,
        -0.055185713,
        -0.10828243,
        -0.03852848,
        0.0056200344,
        0.08000694,
        -0.0064037573,
        -0.08076505,
        0.019398041,
        -0.08192955,
        -0.04676179,
        -0.042482834,
        -0.029019473,
        0.04022647,
        -0.076677956,
        0.057571605,
        -0.018785417,
        -0.040103175,
        -0.015386794,
        0.09070687,
        -0.046670806,
        -0.0041884515,
        0.025041938,
        0.08346705,
        -0.072358064,
        -0.031132871,
        0.06896686,
        0.014795252,
        -0.017799387,
        0.05480383,
        -0.063302346,
        -0.002001896,
        0.04253327,
        0.0040011867,
        -0.025362466,
        -0.021228822,
        -0.06978255,
        0.033685625,
        -0.028204868,
        0.015543864,
        -0.0017123448,
        0.0032268008,
        -0.058024492,
        -0.021608707,
        0.047872555,
        -0.033542063,
        -0.000978261,
        -0.07157343,
        0.0046185795,
        -0.02619723,
        0.045062587,
        -0.070992894,
        -0.08228319,
        -0.58894205,
        -0.039383892,
        -0.032682735,
        -0.019837093,
        -0.15762077,
        -0.020744927,
        0.032791913,
        -0.0021699858,
        0.08042097,
        -0.012945844,
        0.013290493,
        -0.05057716,
        0.0042859367,
        0.07304661,
        0.015986275,
        0.08180095,
        -0.007877518
      ]
    },
    {
      "id": "worldsfair-speaker-153",
      "conference": "worldsfair",
      "name": "Wallon Walusayi",
      "role": "",
      "company": "Qodo",
      "twitter": "",
      "talks": [
        {
          "title": "AI Engineering & Governance 2026 Trends",
          "description": "AI Engineering & Governance 2026 Trends",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "AI Engineering & Governance 2026 Trends",
          "description": "AI Engineering & Governance 2026 Trends",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.014359668,
        0.0144472895,
        0.06797428,
        0.044070557,
        -0.0008734775,
        0.013383897,
        -0.04624737,
        0.055116434,
        -0.052235615,
        -0.2671198,
        -0.038568884,
        -0.040825512,
        0.003183091,
        -0.014569338,
        0.011143214,
        0.016033366,
        0.090335995,
        -0.029479258,
        0.006774713,
        -0.043935683,
        -0.049962573,
        -0.04987656,
        -0.04769319,
        0.022955872,
        -0.103674494,
        0.08775559,
        -0.022982048,
        0.003945098,
        -0.059566043,
        0.45585087,
        -0.051488236,
        -0.012725386,
        -0.01463935,
        -0.03226377,
        0.04893537,
        0.0044285236,
        -0.006891683,
        0.00689141,
        0.028940054,
        0.040082283,
        -0.06931328,
        0.028839257,
        -0.09546433,
        0.08771734,
        0.005559192,
        0.026851006,
        -0.01725761,
        -0.013898947,
        -0.013929053,
        -0.033430748,
        -0.060019605,
        0.000055275403,
        0.12438271,
        -0.01050415,
        0.027554628,
        -0.016995702,
        -0.055708863,
        -0.022123747,
        0.09273864,
        -0.07203503,
        -0.09076547,
        -0.040555444,
        -0.061725244,
        0.028782466,
        -0.07156997,
        0.014564191,
        0.037028015,
        -0.052625507,
        -0.024124492,
        -0.060670305,
        -0.03438662,
        -0.0122318985,
        -0.023152448,
        0.018202081,
        -0.08115088,
        -0.103696674,
        -0.087219805,
        0.04555008,
        -0.03553302,
        -0.048622392,
        0.08018163,
        0.07383999,
        -0.014473305,
        -0.022086535,
        0.05025328,
        0.032832485,
        -0.106570534,
        0.0011146584,
        0.013305312,
        0.027043566,
        -0.0014702023,
        -0.04345575,
        -0.057584275,
        0.09035006,
        0.037983164,
        0.02993151,
        -0.095761746,
        0.005457953,
        -0.060851958,
        0.047755744,
        -0.06325031,
        -0.039883707,
        0.045388307,
        -0.017902238,
        0.028979285,
        -0.021306608,
        -0.0015240709,
        -0.045745928,
        0.043436527,
        0.023088904,
        -0.017466553,
        -0.6225839,
        0.02618754,
        -0.0499783,
        0.01924969,
        -0.14994065,
        0.02720827,
        0.0050483183,
        0.07640142,
        0.116988935,
        -0.0028678018,
        0.09829239,
        -0.008644963,
        -0.0126442965,
        0.051636294,
        0.059104722,
        0.05906165,
        -0.032223526
      ]
    },
    {
      "id": "worldsfair-speaker-154",
      "conference": "worldsfair",
      "name": "Gabriel Spencer-Harper",
      "role": "CEO",
      "company": "Meticulous",
      "twitter": "",
      "talks": [
        {
          "title": "Why AI Didn't Actually Make You Ship Faster",
          "description": "AI generates code faster than humans can review and verify it, and most engineering teams adopting codegen have hit the same wall: verification.\nIn this session, Gabriel (CEO of Meticulous) breaks down why assertion-based testing has a structural ceiling that AI codegen has made impossible to ignore, what exhaustive verification actually requires technically (behavior capture, determinism, and backend isolation), and why the teams solving this now are the ones who will ship at the speed AI enables.\nThe talk includes case studies from LaunchDarkly, which saw an 80% reduction in major frontend incidents after rollout, and Notion, which deployed verification infrastructure across every engineer on every PR to confidently adopt AI-generated code at scale.",
          "day": "Day 3 — Session Day 2",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.009012247,
        0.08193576,
        0.02614679,
        0.034044616,
        -0.01793443,
        -0.0047627534,
        -0.10556349,
        -0.027769292,
        0.03555714,
        -0.21267235,
        -0.016453564,
        0.0065741884,
        0.008291416,
        -0.07901023,
        -0.035626374,
        0.017551592,
        0.0710325,
        -0.079406336,
        -0.018889204,
        -0.058102183,
        0.0062595587,
        -0.007385716,
        -0.019315103,
        0.08123228,
        -0.029983774,
        0.087308675,
        0.057482768,
        0.047191698,
        -0.12403379,
        0.48413813,
        -0.068306945,
        0.007825844,
        0.034977492,
        -0.062206186,
        0.063299194,
        -0.05015708,
        -0.020031044,
        0.0032839677,
        -0.0035355054,
        0.099572055,
        -0.031130735,
        0.044469927,
        -0.041943777,
        0.044367764,
        -0.031295784,
        -0.05178908,
        -0.04815922,
        -0.026581382,
        -0.004534815,
        -0.050658427,
        -0.03227215,
        0.03442816,
        0.03957475,
        0.008404191,
        0.0241682,
        0.056335915,
        -0.026915131,
        -0.06496311,
        0.08679904,
        -0.04987053,
        -0.0928561,
        -0.031700976,
        0.030271089,
        0.012283358,
        0.06082432,
        -0.09582064,
        0.056247845,
        0.009588301,
        -0.02260081,
        -0.05680194,
        0.03763796,
        0.01015294,
        -0.089501716,
        0.014388557,
        -0.10678913,
        -0.087855406,
        -0.04765957,
        0.07574572,
        -0.009295598,
        0.028876523,
        0.012219732,
        0.021769265,
        -0.04994023,
        -0.032348093,
        0.044594277,
        -0.03083447,
        -0.052615434,
        0.02027249,
        -0.032335155,
        0.09047775,
        -0.01861592,
        -0.0392259,
        0.002325544,
        0.022355424,
        0.00855198,
        -0.0006369741,
        -0.067174956,
        0.013383863,
        -0.01222862,
        0.021445153,
        -0.021775655,
        -0.016926456,
        0.02213919,
        -0.014340136,
        -0.036630664,
        -0.031444747,
        -0.012615698,
        0.012985795,
        0.03657567,
        -0.004404587,
        0.0012857879,
        -0.630304,
        0.028796004,
        0.00011221341,
        -0.015104893,
        -0.16101544,
        0.006002181,
        -0.018787814,
        0.04837333,
        0.09044087,
        0.041119985,
        0.10275418,
        -0.024339896,
        0.094968654,
        0.022511967,
        0.07759132,
        0.09057498,
        -0.018324299
      ]
    },
    {
      "id": "worldsfair-speaker-155",
      "conference": "worldsfair",
      "name": "Kai Rikhye",
      "role": "",
      "company": "Radical AI",
      "twitter": "",
      "talks": [
        {
          "title": "No Sleep Til Breakthrough: Building a Long Horizon Agentic Scientist",
          "description": "At Radical AI we're building a self-driving lab for autonomous materials discovery, powered by a long-horizon agentic scientist that manages experimental campaigns spanning weeks. Before we built this technology, materials discovery required several years of hundreds of PhD-level specialists in national labs. Now, in a 16-week campaign we're able to discover 300 novel materials designed for specific performance goals. Building an agentic scientist involves challenges beyond the typical AI assistant. How do you design a system that can reliably and safely operate a physical laboratory containing heavy machinery? How do you manage context across thousands of papers read and hundreds of simulations run? How do you evaluate performance when sample sizes are relatively small and feedback loops are long? In this talk, I'll share learnings from building the AI layer powering a forthcoming 45,000 square foot lab supported by a $55 million seed round.",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.013713404,
        0.030142013,
        0.024147773,
        0.05931097,
        0.0010361545,
        0.007953753,
        -0.105658695,
        0.028805517,
        0.010958027,
        -0.24778962,
        -0.007841892,
        -0.004333414,
        0.034697022,
        0.005420117,
        -0.08841035,
        0.008354011,
        0.08071419,
        -0.08397291,
        -0.0008783389,
        -0.022124834,
        0.008098414,
        -0.07775492,
        -0.019360602,
        0.055676274,
        -0.041497152,
        0.025034396,
        0.05358997,
        0.069809824,
        -0.045384157,
        0.45503515,
        -0.08415946,
        -0.00681268,
        -0.0052484134,
        -0.0030587371,
        0.06528566,
        -0.043589257,
        -0.011626112,
        0.013304172,
        0.049857114,
        0.09621995,
        -0.05620282,
        0.032476276,
        -0.013972146,
        -0.003971544,
        -0.07817542,
        -0.0026763289,
        -0.057520065,
        -0.027988996,
        -0.049004383,
        -0.003503505,
        -0.08114024,
        0.0377209,
        0.009521897,
        0.0023544333,
        0.05634773,
        0.10982012,
        -0.005182105,
        -0.06520718,
        0.09208463,
        -0.04014008,
        -0.067887,
        0.042762503,
        0.012004609,
        -0.029058171,
        -0.0035151967,
        -0.07553436,
        0.08953468,
        -0.01990222,
        0.019317077,
        -0.07523472,
        0.0014878222,
        -0.0072121983,
        -0.10410046,
        0.05146853,
        -0.07904191,
        -0.08372189,
        -0.046409123,
        0.085446045,
        -0.0028099264,
        0.017246846,
        0.013337119,
        0.055208072,
        -0.07766812,
        -0.053751003,
        -0.007812514,
        -0.051895633,
        -0.037201084,
        0.017023886,
        0.00916826,
        0.11905614,
        0.036540646,
        0.045820896,
        -0.08651601,
        -0.021651363,
        -0.060935203,
        -0.026150078,
        -0.0030372427,
        0.058292355,
        -0.005796249,
        0.040896505,
        0.010508987,
        -0.03410002,
        0.068116456,
        -0.009783943,
        -0.032834653,
        -0.010428337,
        0.027709672,
        0.026915662,
        0.08093943,
        -0.0628407,
        -0.05023668,
        -0.6254788,
        -0.03292906,
        0.008060232,
        -0.059635147,
        -0.16805232,
        0.048429564,
        -0.048328605,
        0.04389437,
        0.06747209,
        -0.091510676,
        0.07373084,
        0.002060492,
        0.014002025,
        -0.005692707,
        0.025533265,
        0.033426587,
        -0.017205248
      ]
    },
    {
      "id": "worldsfair-speaker-156",
      "conference": "worldsfair",
      "name": "Ronak Malde",
      "role": "",
      "company": "Trajectory.ai",
      "twitter": "",
      "talks": [
        {
          "title": "Scaling up Continual Learning",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.07903772,
        0.011916979,
        0.05198863,
        0.0614566,
        0.016378246,
        -0.033861227,
        -0.027463084,
        0.025889205,
        0.028685201,
        -0.21655227,
        0.0027870012,
        -0.027372584,
        0.018082023,
        0.036976077,
        -0.070411935,
        -0.046791513,
        0.1018923,
        -0.04265741,
        -0.051193412,
        -0.07722797,
        -0.04177862,
        -0.049333,
        -0.064874046,
        0.0327159,
        -0.040350415,
        0.08751095,
        -0.025224818,
        0.02663682,
        -0.12658034,
        0.47930318,
        -0.10176713,
        -0.11918868,
        0.02692532,
        0.00045152853,
        0.023595192,
        -0.05323117,
        0.03909053,
        0.04405318,
        -0.029887639,
        0.067213334,
        0.0044862214,
        -0.0053004744,
        -0.02478535,
        0.031244054,
        -0.009949584,
        0.016397985,
        -0.09277004,
        -0.006110211,
        -0.027106361,
        -0.021958089,
        -0.103803985,
        0.038042538,
        0.0071265856,
        -0.03823082,
        0.021298539,
        0.08818157,
        -0.043306127,
        -0.013135031,
        0.045236092,
        -0.07448214,
        -0.08502163,
        -0.013661364,
        0.057557307,
        0.017274829,
        -0.049116954,
        -0.044335086,
        0.064282805,
        -0.06309782,
        0.0017310235,
        0.012635569,
        0.051681325,
        -0.012762572,
        -0.04199522,
        0.04064339,
        -0.082280375,
        -0.018246021,
        -0.03384925,
        0.12644245,
        -0.008725434,
        0.028653326,
        -0.004923862,
        0.030891119,
        -0.06369341,
        -0.03566165,
        -0.007841715,
        -0.039894816,
        -0.07219417,
        0.020938084,
        0.0110306805,
        0.03581601,
        0.0466583,
        -0.047507823,
        -0.011381119,
        -0.047033757,
        -0.032042075,
        0.006829502,
        -0.016880682,
        0.03094365,
        -0.040274244,
        0.017279772,
        -0.060803466,
        -0.04984385,
        0.06847155,
        -0.032761518,
        -0.017267577,
        -0.08870212,
        0.002811,
        -0.03361524,
        0.014218062,
        0.008293341,
        -0.08863401,
        -0.62318075,
        0.007023104,
        0.012700043,
        -0.053375244,
        -0.19044086,
        -0.011057645,
        -0.057491962,
        0.0010422926,
        0.10408816,
        -0.014147885,
        0.03230447,
        -0.03603085,
        0.016170254,
        -0.043269556,
        -0.029596929,
        0.03654772,
        0.004090992
      ]
    },
    {
      "id": "worldsfair-speaker-157",
      "conference": "worldsfair",
      "name": "Jason Lopatecki",
      "role": "Co-Founder & CEO",
      "company": "Arize AI",
      "twitter": "",
      "talks": [
        {
          "title": "From Alert to PR: Building an Autonomous Agent Debugger",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.110785,
        0.041883312,
        0.015068538,
        0.03901438,
        0.017671634,
        -0.0064734207,
        -0.054323684,
        -0.030635437,
        0.11450547,
        -0.19485307,
        -0.041951083,
        -0.092127934,
        0.015614202,
        -0.0071909665,
        -0.044979617,
        0.041093014,
        0.09107272,
        -0.05263396,
        -0.027829997,
        -0.10544881,
        -0.07461472,
        -0.031053782,
        -0.021243645,
        0.06536299,
        0.02936361,
        0.17383225,
        0.026624499,
        0.044572417,
        -0.12073056,
        0.46441874,
        -0.08249268,
        -0.0076865195,
        -0.010839098,
        -0.03570435,
        0.026507087,
        -0.04309967,
        -0.00034073542,
        -0.0071457187,
        0.0095011555,
        0.04701673,
        -0.020377563,
        0.019014666,
        0.021709321,
        0.04079259,
        -0.033889208,
        0.0067156106,
        -0.103678405,
        -0.029666482,
        -0.021405656,
        -0.007927463,
        -0.014683365,
        0.010360425,
        -0.00398675,
        -0.015457168,
        0.04331245,
        0.038827933,
        0.0038615945,
        -0.082615,
        0.012005756,
        -0.068887755,
        -0.106887385,
        -0.027082719,
        -0.021978918,
        0.08310194,
        0.028664364,
        -0.07711936,
        0.037389435,
        -0.05303864,
        0.015441655,
        -0.047081932,
        0.038943067,
        0.048325043,
        -0.09711081,
        0.023459332,
        -0.12280818,
        -0.07082187,
        -0.012434508,
        0.11013877,
        -0.07333546,
        -0.048401378,
        0.058642205,
        0.06375688,
        -0.029751122,
        -0.058498282,
        0.070078716,
        -0.0148868365,
        -0.08317998,
        0.058166232,
        0.020453306,
        0.049780767,
        0.052868284,
        -0.057975274,
        0.021482097,
        0.03707377,
        -0.05129131,
        -0.0024883084,
        -0.06922855,
        0.08671268,
        0.01628511,
        -0.026964856,
        -0.021400752,
        -0.004632819,
        0.09377959,
        -0.018488863,
        -0.025185816,
        -0.048212662,
        0.013445879,
        -0.0012782568,
        0.011108409,
        0.005337283,
        -0.07841047,
        -0.6012558,
        -0.019951053,
        -0.020939728,
        -0.057374105,
        -0.16756363,
        0.04826966,
        0.0033426716,
        0.029714081,
        0.0518567,
        0.012149641,
        0.059845787,
        0.016523408,
        0.052975565,
        0.011388037,
        0.05608553,
        0.038786367,
        -0.0034995028
      ]
    },
    {
      "id": "worldsfair-speaker-158",
      "conference": "worldsfair",
      "name": "Max Drake",
      "role": "",
      "company": "tldraw",
      "twitter": "https://x.com/max__drake",
      "talks": [
        {
          "title": "The Spatial Harness: Bringing Agents to the Canvas",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.037951693,
        0.023222854,
        0.07187198,
        0.041708663,
        0.05251434,
        -0.05858745,
        -0.02036756,
        0.009627874,
        -0.035557322,
        -0.25807098,
        -0.041643247,
        -0.08182712,
        0.00033902706,
        0.06319313,
        -0.025910014,
        -0.02720416,
        0.10744413,
        -0.046921246,
        -0.0662934,
        -0.09797784,
        -0.044944648,
        -0.069619805,
        -0.06569903,
        -0.009625196,
        -0.06809486,
        0.08870856,
        0.006161759,
        0.04422594,
        -0.13590409,
        0.4567034,
        -0.118390456,
        0.013325858,
        -0.008059087,
        0.026020385,
        0.109033905,
        -0.1003805,
        -0.03132482,
        0.0006210146,
        -0.013866437,
        0.039625693,
        -0.005575033,
        -0.029546395,
        -0.021047862,
        0.00083934265,
        -0.077101715,
        0.09449094,
        -0.05914204,
        -0.032621697,
        -0.0612286,
        -0.02261911,
        -0.06560956,
        0.053705815,
        0.037585206,
        -0.068489455,
        -0.026438786,
        -0.002742588,
        -0.031163987,
        -0.09927822,
        0.055207383,
        -0.039701674,
        -0.06363626,
        0.005160521,
        -0.037502516,
        0.03204115,
        0.057973944,
        -0.04654108,
        0.04722878,
        -0.056360275,
        -0.0020239719,
        -0.04968142,
        0.05345114,
        -0.00912439,
        -0.030768152,
        0.055924352,
        -0.092162065,
        0.0033204705,
        -0.06583611,
        0.119009644,
        0.010371299,
        0.0032750512,
        0.016882114,
        -0.023071973,
        -0.05853692,
        -0.10821297,
        -0.010934571,
        -0.090622805,
        -0.018930405,
        0.064638965,
        -0.010010152,
        0.002580334,
        0.021687094,
        0.027519716,
        -0.07399572,
        0.013262998,
        -0.025232617,
        -0.010044505,
        -0.033687633,
        0.088379756,
        0.0049158665,
        0.042085312,
        -0.019943217,
        -0.061005034,
        0.06086366,
        -0.058885045,
        -0.066733696,
        -0.045201883,
        0.051052984,
        -0.03620757,
        0.04378953,
        -0.063590705,
        -0.009222321,
        -0.60634893,
        -0.07358302,
        0.013171117,
        -0.017957333,
        -0.12643431,
        -0.049887516,
        0.03684939,
        -0.015648436,
        0.027524713,
        -0.0028331552,
        0.014425518,
        -0.03459107,
        0.032143295,
        0.0246979,
        0.035692308,
        0.009097032,
        0.010416728
      ]
    },
    {
      "id": "worldsfair-speaker-159",
      "conference": "worldsfair",
      "name": "Kevin Madura",
      "role": "",
      "company": "AlixPartners",
      "twitter": "https://x.com/kmad",
      "talks": [
        {
          "title": "It’s Tokens All The Way Down: How RLMs are Different",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.06080428,
        0.040422637,
        0.014237435,
        0.095797785,
        0.09645819,
        -0.067319915,
        -0.07198174,
        0.013935833,
        0.028598925,
        -0.19954818,
        -0.044722825,
        -0.044142198,
        0.011226115,
        0.02880114,
        -0.051785655,
        -0.0042978055,
        0.12452522,
        -0.1663628,
        -0.06318181,
        -0.08217467,
        -0.010118956,
        -0.0819219,
        -0.02295288,
        -0.021330493,
        -0.06745123,
        0.075469196,
        0.027064027,
        0.0058751595,
        -0.0025151144,
        0.49425426,
        -0.07150282,
        -0.010241691,
        -0.03366517,
        0.05036195,
        0.021430317,
        -0.03706616,
        -0.038696557,
        0.011004456,
        -0.016271904,
        0.13496397,
        -0.03224399,
        0.03564796,
        -0.09479151,
        0.028099734,
        -0.06325557,
        -0.033938568,
        -0.016026387,
        -0.083292,
        -0.09354193,
        -0.06284466,
        -0.08918767,
        0.03478502,
        0.03066252,
        0.0035059913,
        0.032138336,
        0.018613989,
        -0.022110585,
        0.0118242465,
        0.08050998,
        -0.039529,
        -0.03863861,
        0.0722658,
        0.00243326,
        0.0090359105,
        -0.009794948,
        0.031336665,
        0.05062555,
        0.011892894,
        -0.022213276,
        0.04145995,
        0.019489773,
        -0.03131762,
        -0.044153746,
        -0.020006469,
        -0.022169584,
        -0.05396593,
        -0.041714177,
        0.099638216,
        -0.065525934,
        0.01726919,
        0.07419234,
        0.027048355,
        -0.08555043,
        -0.06609848,
        0.017225817,
        -0.0139361415,
        -0.04727807,
        0.03051843,
        0.030608933,
        0.10596042,
        0.010437766,
        0.015892465,
        -0.0072091497,
        -0.07779132,
        -0.09338337,
        -0.014523835,
        -0.11571843,
        0.037658244,
        -0.062006373,
        -0.021402858,
        -0.026353,
        0.033803757,
        0.07278983,
        0.03220097,
        0.008506518,
        0.006620961,
        -0.037238657,
        -0.06106013,
        0.087661475,
        -0.032883003,
        0.014296281,
        -0.58888257,
        0.009288281,
        -0.008277915,
        0.00710128,
        -0.17386271,
        0.017723037,
        -0.021119067,
        -0.032259803,
        0.021973122,
        -0.04374685,
        -0.03898252,
        -0.025597867,
        -0.014664865,
        -0.003256774,
        -0.034201033,
        0.028249338,
        0.010220789
      ]
    },
    {
      "id": "worldsfair-speaker-160",
      "conference": "worldsfair",
      "name": "Andrew Garvin",
      "role": "",
      "company": "Metronome",
      "twitter": "",
      "talks": [
        {
          "title": "How to avoid disaster when vibe-coding a billing engine",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.056935042,
        0.06798071,
        0.006698919,
        0.013726275,
        -0.008216312,
        -0.04419293,
        -0.08984472,
        -0.0031659366,
        0.059870653,
        -0.19600733,
        -0.0758485,
        -0.07208293,
        0.04612946,
        0.049357064,
        -0.020153305,
        -0.004412718,
        0.10063004,
        -0.09676654,
        -0.03241103,
        -0.110435456,
        -0.018502224,
        -0.11605241,
        -0.05884115,
        -0.0018221173,
        -0.03508093,
        0.14797321,
        -0.036381133,
        0.072465636,
        -0.09984443,
        0.49005404,
        -0.060229816,
        0.022513602,
        0.0025253845,
        0.014589809,
        0.08215944,
        -0.07996779,
        0.042604726,
        -0.007996947,
        -0.052753814,
        0.09800091,
        0.005058707,
        -0.043819305,
        -0.10835035,
        -0.03388526,
        0.044616174,
        0.02010289,
        -0.056658525,
        -0.07039662,
        0.055609643,
        -0.07792811,
        -0.06373825,
        0.0036667967,
        0.017534282,
        0.0030882256,
        -0.0072051515,
        0.0029065688,
        0.020229151,
        -0.035648428,
        0.11144962,
        -0.024937514,
        -0.052029707,
        0.026705423,
        -0.00016029457,
        -0.00936839,
        -0.0022528123,
        -0.039425027,
        0.097206526,
        0.013050571,
        0.020292709,
        -0.011279492,
        0.015821286,
        0.033718128,
        -0.06025003,
        0.00840826,
        -0.011553292,
        -0.07077915,
        0.014773184,
        0.04136667,
        -0.057978194,
        0.09379695,
        0.07517519,
        -0.060285427,
        -0.032955267,
        -0.07320971,
        -0.035155114,
        -0.03616529,
        0.0049814503,
        0.0502521,
        -0.05032515,
        0.039969224,
        -0.018499076,
        -0.074156925,
        0.04708659,
        0.000654477,
        -0.00057964283,
        0.041885626,
        -0.035354935,
        0.049948815,
        -0.016930817,
        -0.03167627,
        -0.04066106,
        -0.03728033,
        0.042945072,
        -0.012432831,
        -0.049328987,
        -0.039920468,
        -0.062046576,
        0.004945269,
        0.03153327,
        -0.042457204,
        -0.009008351,
        -0.60970855,
        -0.009417669,
        -0.029430222,
        -0.013285122,
        -0.1464674,
        -0.035088394,
        0.046747424,
        -0.03735149,
        0.05653747,
        0.043981716,
        -0.003373387,
        -0.008261363,
        -0.041864954,
        0.0103969285,
        0.0012454292,
        0.08149422,
        0.022287184
      ]
    },
    {
      "id": "worldsfair-speaker-161",
      "conference": "worldsfair",
      "name": "Vlad Luzin",
      "role": "",
      "company": "Band.ai",
      "twitter": "",
      "talks": [
        {
          "title": "The Next Inflection Point: ChatGPT, Claude Code, OpenClaw — what's next?",
          "description": "Every so often a new capability comes along that changes the way we see everything that follows. ChatGPT was one: overnight, the whole world was talking to a model. Claude Code was another: the model stopped answering questions and started writing and running real software, and the way engineers work changed under their feet. Then came OpenClaw — an agent that broke out of the terminal and ran loose across people's messaging apps, shells, and files — and people everywhere reorganized their daily lives around it almost overnight. That's what an inflection point is: a capability that doesn't just add something new, but resets what everybody thought was possible — and there's no going back. So what comes next? Three of these in two years — is that a pattern we can read, or just hindsight? This talk lines them up, looks for what they share, and makes the case for what the next world-changing capability is.",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.038078606,
        -0.037950896,
        0.03118837,
        0.022759518,
        -0.011979054,
        -0.041373387,
        -0.014356155,
        0.02687712,
        0.065947294,
        -0.25859088,
        -0.070933565,
        -0.009360247,
        0.043397322,
        -0.06365042,
        -0.0048120455,
        0.02199306,
        0.15677801,
        -0.054036085,
        -0.099801734,
        -0.076053426,
        -0.053457815,
        -0.07315311,
        -0.045416288,
        0.018790992,
        -0.034516998,
        0.11018642,
        -0.022912098,
        0.057011295,
        -0.1171883,
        0.45667422,
        -0.06433492,
        -0.036262035,
        -0.0024412237,
        -0.0035630113,
        0.06028024,
        -0.11928851,
        -0.044570666,
        -0.015319289,
        -0.01808353,
        0.076291375,
        0.016482946,
        -0.025044998,
        -0.050802935,
        -0.004773005,
        -0.015581165,
        0.023175001,
        -0.0720892,
        0.0069527472,
        0.005712883,
        0.017468218,
        -0.03813562,
        0.035941884,
        0.09348037,
        -0.020583691,
        0.050471038,
        0.011992071,
        0.05788267,
        -0.08116247,
        0.033111244,
        -0.043487504,
        -0.048970077,
        -0.052097593,
        -0.03005748,
        0.07937266,
        -0.02106107,
        -0.02458331,
        -0.013750424,
        -0.11112322,
        -0.013327094,
        0.010558171,
        -0.015790863,
        0.01593717,
        -0.009524538,
        0.021574391,
        0.017859522,
        -0.033633824,
        -0.05137047,
        0.021501025,
        -0.041989774,
        -0.077753164,
        0.050902996,
        0.042414732,
        -0.081321,
        0.021310356,
        0.029424489,
        -0.0019093187,
        0.01512903,
        0.024509551,
        -0.07639021,
        0.028456949,
        0.059672788,
        -0.08067976,
        -0.023159835,
        0.032683253,
        -0.12992735,
        0.035555974,
        -0.05751464,
        0.05699345,
        -0.011088508,
        -0.0013761788,
        0.00198188,
        -0.0479236,
        0.02389604,
        0.028448723,
        0.035048846,
        -0.051243316,
        -0.019755004,
        -0.007706558,
        0.03909363,
        0.0068485015,
        -0.036427096,
        -0.6080684,
        0.017601391,
        -0.007327806,
        0.021108447,
        -0.16070703,
        0.032457072,
        0.038274977,
        0.0800777,
        0.14242984,
        -0.021515483,
        0.0698035,
        -0.04749916,
        -0.025640113,
        0.028637907,
        0.036144882,
        0.07009447,
        0.0034807767
      ]
    },
    {
      "id": "worldsfair-speaker-162",
      "conference": "worldsfair",
      "name": "Justin Reock",
      "role": "Deputy CTO",
      "company": "DX",
      "twitter": "",
      "talks": [
        {
          "title": "AI-Assisted Engineering: 5 Trends We're Seeing From 500+ Organizations",
          "description": "AI is reshaping how engineers work but what does that actually look like at scale? Drawing on data and patterns from more than 500 organizations, we break down the five most significant trends emerging in AI-assisted engineering today.\n\nThis fast-paced theater session cuts through the hype to deliver concrete, evidence-based insights that engineering leaders can act on immediately.\n\nKey takeaways:\n\nDiscover the top 5 AI-assisted engineering trends observed across 500+ organizations\n\nUnderstand how leading teams are integrating AI into their engineering workflows\n\nLeave with actionable strategies to apply at your organization",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        },
        {
          "title": "The state of AI in software development: Insights across 400+ organizations",
          "description": "Headlines claim AI is transforming software engineering overnight. Across more than 400 engineering organizations, we see patterns that challenge the hype and reveal what's really working, and what isn't, when AI enters the software development lifecycle.\n\nIn this talk, Justin Reock, Deputy CTO at DX, will share a data-driven \"state of the union\" on AI in engineering, grounded in both quantitative data from thousands of developers and on-the-ground observations.\nYou'll learn:\n\nThe current impact of AI, from benchmarks on the percentage of code authored, team PR throughput, and time savings\n\nWhere AI adoption is creating real gains in throughput, and whether it introduces tradeoffs for quality and maintainability\n\nInsights and trends, including whether junior or senior developers are seeing bigger gains, the impact of structured rollouts, which tools are having the most impact, and the evolving definition of \"developer\"\n\nThe session will conclude with a practical framework for measuring AI's impact, helping leaders cut through hype and understand the impact AI is having in their own organizations.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.059936147,
        0.044066302,
        -0.01376859,
        -0.014815463,
        0.056832004,
        -0.021802818,
        -0.079700634,
        0.023350207,
        0.045367163,
        -0.22327338,
        -0.04113982,
        -0.016950816,
        0.02095112,
        -0.036705904,
        -0.06935282,
        -0.017039059,
        0.11583846,
        -0.03419813,
        -0.10844872,
        -0.0503152,
        -0.017937891,
        -0.06541132,
        -0.04578396,
        0.04720368,
        -0.04880841,
        0.10403514,
        0.009343601,
        0.06981607,
        -0.08744803,
        0.43468633,
        -0.081641264,
        0.025418017,
        0.043898743,
        -0.018181553,
        0.02602445,
        -0.06870952,
        -0.055581734,
        -0.04065373,
        -0.007923703,
        0.09832091,
        -0.0029873445,
        0.026625523,
        -0.0780986,
        0.03216883,
        -0.012511573,
        -0.02318619,
        -0.04671355,
        -0.06706264,
        0.008643813,
        -0.09393757,
        -0.061530884,
        -0.04763528,
        0.034826357,
        -0.0473367,
        0.030048419,
        0.058687564,
        -0.06047786,
        -0.15243512,
        0.07728295,
        -0.0802012,
        -0.15436752,
        -0.038533926,
        0.019345516,
        0.041702703,
        -0.00551772,
        -0.04299971,
        -0.010121551,
        -0.046195544,
        -0.059287786,
        -0.08508844,
        0.0039092414,
        0.075435266,
        -0.11108348,
        -0.0034362485,
        -0.052748553,
        -0.104454696,
        -0.099382535,
        0.0823497,
        -0.04295771,
        -0.0009733638,
        -0.035992365,
        0.041621678,
        -0.07474063,
        -0.036122095,
        -0.00064911606,
        0.0020304155,
        0.0047332654,
        0.027731722,
        -0.09220617,
        0.07485753,
        0.024473418,
        -0.06154009,
        -0.03323311,
        0.10089849,
        -0.083896026,
        0.0043656942,
        -0.016282203,
        0.0492543,
        0.029401826,
        0.003287929,
        -0.015869617,
        -0.08013354,
        0.02389507,
        0.0028232923,
        -0.007970086,
        -0.08915421,
        0.021071194,
        -0.008458482,
        0.05803065,
        -0.001132166,
        0.002777442,
        -0.582965,
        0.022909157,
        0.016050907,
        0.041799113,
        -0.12785645,
        0.050507277,
        -0.02189505,
        0.037910324,
        0.09467251,
        -0.008096408,
        0.056893956,
        -0.12256393,
        0.01988629,
        -0.014097288,
        0.095732175,
        -0.00006786883,
        -0.017986916
      ]
    },
    {
      "id": "worldsfair-speaker-163",
      "conference": "worldsfair",
      "name": "Nixon Dinh",
      "role": "",
      "company": "PayPal",
      "twitter": "",
      "talks": [
        {
          "title": "The Death of Keyword Search and the Rise of Agent-Readable Catalogs",
          "description": "As search shifts from classic keyword matching to more conversational experiences, product data quality becomes critical to LLM-powered retrieval. At PayPal, we tested how enriching traditional catalog data could help AI systems better find, understand, and rank products across large-scale commerce catalogs. We built a RAG-based AI judge to compare enrichment approaches and identify five patterns that consistently improved AI discovery results.In this talk, we'll share the evaluation framework, key lessons, and a practical approach for preparing enterprise data for conversational and agentic search.",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.031325065,
        0.09121756,
        -0.044902477,
        0.010720379,
        0.02632822,
        -0.07463188,
        -0.12287932,
        0.02183972,
        0.0064617204,
        -0.21992391,
        0.016739156,
        -0.055054534,
        0.04433664,
        0.007946504,
        -0.041158944,
        0.032634262,
        0.15825549,
        -0.08605686,
        -0.10162434,
        -0.04806566,
        -0.060371693,
        -0.0148103805,
        -0.001382333,
        0.01398844,
        -0.040176146,
        0.03148298,
        0.06587374,
        0.081335984,
        -0.042956345,
        0.4711224,
        -0.038108222,
        0.0035155483,
        -0.019054933,
        -0.024893573,
        -0.012195843,
        -0.03990496,
        0.027849624,
        0.012477865,
        0.04469202,
        0.035211325,
        0.024941184,
        0.026823346,
        -0.05865052,
        0.03530161,
        -0.03064706,
        -0.0859733,
        -0.08383432,
        0.0138260545,
        0.037483502,
        -0.0043230955,
        -0.060051795,
        0.0043912674,
        0.06592759,
        -0.010078341,
        0.048384424,
        0.049655594,
        0.017503986,
        -0.05815818,
        0.11346387,
        -0.056847423,
        -0.1134369,
        0.033743832,
        0.005263328,
        -0.031213047,
        -0.020143127,
        -0.084811606,
        0.06799209,
        -0.06349449,
        0.03001735,
        -0.015891455,
        -0.050293706,
        -0.00068815605,
        -0.066546805,
        0.011966545,
        -0.05761779,
        -0.15531194,
        -0.064123556,
        -0.006899782,
        -0.040813986,
        0.055548634,
        0.007983671,
        0.039759025,
        -0.051772054,
        -0.030648306,
        0.045681898,
        0.0026466763,
        -0.015028248,
        0.049273882,
        -0.06146016,
        0.12581281,
        -0.037102047,
        -0.023803512,
        0.032415956,
        -0.02772009,
        -0.017409602,
        0.002105391,
        0.01563407,
        0.06457301,
        -0.014766834,
        0.07043342,
        -0.01756309,
        -0.047335867,
        0.04230536,
        -0.022715462,
        0.01106075,
        -0.025601123,
        -0.007927597,
        -0.020116745,
        0.06606703,
        0.036230497,
        -0.061758406,
        -0.6115921,
        0.0155858975,
        -0.007241656,
        0.019238086,
        -0.15519023,
        0.08520674,
        0.012207488,
        -0.024125688,
        0.060628034,
        -0.020108407,
        0.047547683,
        -0.026845315,
        -0.067813486,
        0.018612467,
        0.027133973,
        0.050156426,
        0.053748548
      ]
    },
    {
      "id": "worldsfair-speaker-164",
      "conference": "worldsfair",
      "name": "Matt Lawler",
      "role": "Forward Deployed Engineer Lead",
      "company": "AssemblyAI",
      "twitter": "",
      "talks": [
        {
          "title": "FDE Playbook: Build an AI Support Agent and Give It a Voice",
          "description": "Bio: Matt Lawler leads FDE for Onboarding at AssemblyAI, helping teams ship speech-to-text and voice AI to production, from model selection and architecture through deployment and scale.\nDescription:\nMost support bots can read. Joey can talk back. In this session, AssemblyAI's Forward Deployed Engineer Lead, Matt Lawler, shares how his team built Joey, an AI support agent that increased end-to-end resolution rates from 10% to 75%. He'll walk through the architecture, key lessons learned, and how the team extended Joey into a fully voice-enabled agent.",
          "day": "Day 3 — Session Day 2",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.036958016,
        -0.03888515,
        0.007649346,
        -0.035903223,
        0.010947814,
        -0.006480208,
        -0.05039446,
        0.022753565,
        0.029015098,
        -0.1519779,
        -0.031540766,
        -0.033070557,
        -0.0049588573,
        -0.030863818,
        -0.10691143,
        0.029941661,
        0.07528944,
        -0.08500579,
        -0.03949827,
        -0.092187,
        -0.07426689,
        -0.04187449,
        -0.052858226,
        0.040754333,
        -0.09665278,
        0.11984645,
        -0.011046861,
        0.051164664,
        -0.13547185,
        0.42828053,
        -0.11639768,
        0.05699246,
        -0.047305558,
        0.025516044,
        0.016942425,
        -0.005895793,
        -0.014365265,
        -0.01968229,
        -0.06806898,
        0.026708994,
        0.027776673,
        0.048627958,
        -0.072502665,
        0.016963782,
        0.033734456,
        -0.027172403,
        -0.009989851,
        -0.025041569,
        0.014258959,
        0.0026187322,
        0.016747782,
        0.013543847,
        0.025475226,
        -0.027030753,
        0.02728569,
        0.07136865,
        0.044033762,
        -0.08957865,
        0.08542901,
        -0.09927048,
        -0.10762268,
        0.018279089,
        0.040266536,
        -0.017563181,
        0.026657488,
        -0.09528801,
        0.061397687,
        0.011739716,
        -0.08061867,
        -0.045716967,
        0.05250181,
        0.047536936,
        -0.05840185,
        -0.0001864196,
        -0.11455312,
        -0.16029724,
        -0.032261215,
        0.10565797,
        0.011299707,
        -0.015023896,
        0.044341892,
        0.10313275,
        -0.1228998,
        -0.046359587,
        0.039757267,
        -0.058510095,
        -0.08571256,
        0.041038327,
        -0.024154836,
        0.033729427,
        0.0016616925,
        -0.0068208016,
        -0.044627156,
        -0.014365041,
        -0.05308813,
        -0.018072026,
        -0.06706178,
        0.015575112,
        -0.053706747,
        0.08765665,
        -0.06782344,
        -0.053226095,
        0.046167683,
        -0.059954163,
        -0.037096977,
        -0.063413344,
        -0.021582585,
        -0.003684666,
        0.08638667,
        0.0026737144,
        -0.048054583,
        -0.61652905,
        0.034163423,
        -0.0050888155,
        0.05568548,
        -0.12220331,
        0.045366716,
        0.0066218963,
        -0.021651011,
        0.06315833,
        -0.030805087,
        0.04974409,
        0.0027526699,
        -0.0044722613,
        0.0067565967,
        0.09242465,
        0.09478034,
        0.010235061
      ]
    },
    {
      "id": "worldsfair-speaker-165",
      "conference": "worldsfair",
      "name": "Stefania Druga",
      "role": "Research Scientist",
      "company": "Sakana.ai",
      "twitter": "",
      "talks": [
        {
          "title": "Memory Harnesses for Long-Running Research Agents",
          "description": "At Sakana AI we build agents that run for hundreds of turns to read literature, run experiments, and draft papers. The model rarely breaks. The harness around it is the weak point: the agent contradicts a decision it made 80 turns ago, redoes finished work, or drifts from the question it started on. This is the binding-constraint thesis. For long-horizon tasks, reliability is set as much by the harness as by the model as clearly instantiated in autoresearch recent efforts. This is a field guide to the harness's memory layer. I'll trace a real research agent through its lifecycle, show exactly where context rot and drift set in, and cover the patterns that hold over 100+ turns: three-tier memory, progressive disclosure, recall-first compaction, sub-agent isolation, and architectural memory beyond the vector database. I will show how to measure whether your memory harness actually helps, at the trajectory level, so you stop tuning prompts to fix what's really a state-management bug.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09531152,
        0.03085509,
        0.08546355,
        0.03359303,
        0.018876348,
        -0.026716579,
        -0.018646229,
        -0.0020520012,
        0.044498395,
        -0.22303227,
        -0.022729317,
        -0.048341796,
        0.04991134,
        -0.022402802,
        -0.085471295,
        0.019781189,
        0.11239962,
        -0.043580074,
        -0.043165985,
        -0.06632368,
        0.030260524,
        -0.07457466,
        0.00058928813,
        0.1267698,
        -0.026021777,
        0.104978375,
        0.047211103,
        0.03351247,
        -0.08901759,
        0.43296698,
        -0.099008754,
        -0.015418012,
        0.02109873,
        -0.0363882,
        0.089791514,
        -0.10720083,
        -0.028916301,
        -0.056712218,
        0.012591276,
        0.052937634,
        0.0039245444,
        0.07052355,
        -0.007296824,
        0.051319,
        -0.012405482,
        0.0034752807,
        -0.05021041,
        0.040980972,
        -0.028096927,
        0.015927881,
        -0.018667046,
        0.05928125,
        0.08132736,
        -0.020165874,
        0.04114129,
        0.02146317,
        0.026088217,
        -0.07831267,
        0.09458949,
        -0.016756708,
        -0.102714136,
        0.043393422,
        0.0070267543,
        0.04999554,
        0.03350494,
        -0.07618011,
        0.064893104,
        -0.045297075,
        0.05241676,
        -0.069433205,
        0.044626687,
        0.034884024,
        -0.025814459,
        0.00952628,
        -0.072977215,
        -0.12771605,
        -0.03714184,
        0.058880758,
        -0.007086347,
        0.044003423,
        -0.05580723,
        0.031452466,
        -0.04934815,
        -0.059572484,
        0.0740997,
        -0.015892802,
        -0.048232935,
        0.010446197,
        -0.096735954,
        0.057261337,
        0.06024668,
        -0.020873126,
        -0.116962515,
        -0.013482421,
        -0.084079556,
        -0.004236438,
        -0.058627367,
        0.075087294,
        0.019149229,
        0.034977246,
        -0.010991651,
        -0.027217124,
        0.10023858,
        -0.0449003,
        0.025126666,
        -0.002287904,
        -0.0054438426,
        0.033782475,
        0.08719534,
        0.0015277662,
        -0.020525184,
        -0.5912496,
        -0.054956425,
        0.030378167,
        -0.073493235,
        -0.17385298,
        0.044569135,
        -0.0019867967,
        0.09418711,
        0.07197777,
        -0.076894,
        0.07740871,
        -0.013213518,
        -0.01400405,
        -0.006906314,
        0.028004857,
        0.051545527,
        0.045319203
      ]
    },
    {
      "id": "worldsfair-speaker-166",
      "conference": "worldsfair",
      "name": "Ivan Burazin",
      "role": "CEO",
      "company": "Daytona",
      "twitter": "https://x.com/ivanburazin",
      "talks": [
        {
          "title": "Kubernetes Is Not Your Sandbox",
          "description": "Teams are reaching for Kubernetes to run agent sandboxes, and it's the wrong tool. Kubernetes is built to keep things alive and hold them in a steady state. A sandbox is born, forked, and killed before any of that machinery catches up.\n\nThe mismatch compounds because the sandbox keeps gaining requirements without shedding any. In eighteen months it went from a fast code-snippet runner, to a stateful box for long-running agents, to ten thousand ephemeral environments that fork for RL rollouts and die in under a second. It has to be all of those at once, a contradiction set no orchestrator was designed to hold. \n\nThe cost shows up the moment you measure it. We ran the same 50-action bug-fix trajectory across five stacks and got a 12x spread: 12.9s on the fastest, 161.5s on the slowest. The gap isn't compute, it's lifecycle overhead per action. We name every stack and explain the mechanism behind each number.\nwdyt?",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.019059833,
        0.030463882,
        0.0138781825,
        -0.02410448,
        -0.004202142,
        -0.049639046,
        -0.053122073,
        -0.04870247,
        0.037346974,
        -0.25627607,
        -0.0102853095,
        -0.009240377,
        0.037602067,
        -0.04945464,
        -0.042006567,
        -0.004666266,
        0.11575409,
        -0.10362577,
        -0.10230951,
        -0.12015768,
        -0.070336156,
        -0.057950668,
        -0.023212997,
        0.041673098,
        -0.022620039,
        0.11302363,
        0.039538212,
        0.08035907,
        -0.07499086,
        0.4740151,
        -0.071218155,
        -0.044151228,
        0.08617655,
        -0.04111791,
        0.013565213,
        -0.07590416,
        -0.041681368,
        0.010822036,
        -0.04414724,
        0.065919474,
        0.017807307,
        0.06996084,
        -0.07548852,
        -0.012060111,
        -0.054864027,
        -0.030426962,
        -0.03414302,
        0.0005555982,
        0.010956905,
        -0.004621179,
        -0.03983071,
        0.09801225,
        0.063475944,
        -0.014814097,
        -0.011049823,
        -0.03399155,
        0.058619443,
        -0.0832574,
        0.06758499,
        -0.04136732,
        -0.11868449,
        0.0020543127,
        0.0013196204,
        0.039876115,
        0.0049763694,
        -0.044279415,
        0.021760354,
        -0.0055795214,
        0.02858194,
        0.03519681,
        0.019722924,
        0.08777436,
        -0.0033571657,
        0.018585458,
        -0.08855598,
        -0.06468489,
        0.011364058,
        0.055178892,
        -0.028034266,
        0.022432376,
        -0.027282083,
        0.111461826,
        -0.087678924,
        -0.02863029,
        0.0746704,
        0.022580756,
        -0.05272847,
        0.042205296,
        -0.018400948,
        0.038320124,
        -0.025143957,
        -0.054130398,
        0.0008853616,
        0.03617779,
        -0.07434044,
        0.061191488,
        -0.053422946,
        -0.011488599,
        -0.0022166849,
        0.07416496,
        0.024343515,
        0.035774082,
        -0.0038085228,
        -0.014434951,
        0.021004332,
        -0.09012265,
        0.016532773,
        -0.013329629,
        0.056684542,
        -0.053420864,
        0.008377545,
        -0.57102424,
        -0.031075075,
        -0.058669884,
        -0.031407956,
        -0.18577456,
        0.021610625,
        0.08804051,
        0.06473051,
        0.051422212,
        -0.0059154453,
        0.12054182,
        -0.055383224,
        0.048424907,
        0.046787944,
        0.05334719,
        0.0069744033,
        0.030689964
      ]
    },
    {
      "id": "worldsfair-speaker-167",
      "conference": "worldsfair",
      "name": "Jason Ma",
      "role": "CTO and co-founder",
      "company": "Dyna Robotics",
      "twitter": "",
      "talks": [
        {
          "title": "HOLD — Dyna Robotics / Jason Ma",
          "description": "TBD — Dyna Robotics talk for Robotics & World Models track.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.039397657,
        0.046379976,
        0.00984014,
        -0.0027165613,
        0.05155243,
        -0.041605055,
        -0.023051128,
        -0.007256956,
        0.07773541,
        -0.24744624,
        -0.067623004,
        -0.05014255,
        -0.012309692,
        0.07953872,
        -0.0898299,
        0.0063554207,
        0.079853415,
        -0.05832832,
        -0.032834303,
        -0.05393374,
        -0.00035919508,
        -0.04202656,
        -0.042195138,
        0.058447894,
        -0.0077234227,
        0.079983026,
        0.073986866,
        0.045865014,
        -0.114219636,
        0.42862102,
        -0.096046925,
        -0.016660552,
        0.006792483,
        0.037114277,
        0.10442616,
        -0.06724775,
        -0.07564325,
        -0.100606225,
        0.0137968,
        0.034731966,
        -0.004479263,
        0.040878817,
        0.02330122,
        0.060299713,
        -0.007532397,
        -0.025013514,
        -0.066773795,
        -0.005251025,
        0.021534013,
        0.034808464,
        -0.060067404,
        0.021738864,
        0.052819133,
        0.04424674,
        0.020904196,
        0.023505982,
        0.018513773,
        -0.016249962,
        0.08491694,
        0.022560695,
        -0.11766942,
        0.034601856,
        0.021124654,
        -0.024102354,
        0.04525838,
        -0.08531796,
        0.03695033,
        -0.045291077,
        0.02179156,
        0.015124782,
        0.07610532,
        0.04295364,
        -0.032626126,
        0.03599214,
        -0.0782935,
        -0.09171183,
        -0.030526744,
        0.050700963,
        0.028063873,
        -0.050141476,
        0.026087426,
        0.063837335,
        -0.031908996,
        -0.16809656,
        0.099804886,
        0.0071732732,
        -0.08411726,
        0.05629643,
        -0.012346181,
        0.0543799,
        -0.00062863895,
        -0.039073825,
        -0.09816792,
        -0.015244038,
        -0.096130915,
        0.050265435,
        0.013237894,
        -0.013962522,
        -0.012547614,
        -0.014047925,
        -0.043052044,
        0.048886415,
        0.024579147,
        -0.0029114212,
        -0.027195107,
        -0.0073647457,
        0.050848577,
        -0.025229936,
        0.033264406,
        0.016110724,
        -0.00038510025,
        -0.6292727,
        -0.04414398,
        0.060818583,
        -0.02383294,
        -0.1594908,
        0.026322449,
        -0.05769207,
        0.04466331,
        -0.012386194,
        0.027325429,
        0.072650075,
        -0.002465239,
        0.039494865,
        -0.025996473,
        0.018112019,
        -0.03883898,
        -0.0005769725
      ]
    },
    {
      "id": "worldsfair-speaker-168",
      "conference": "worldsfair",
      "name": "Jack Morris",
      "role": "AI researcher",
      "company": "Cornell / Meta FAIR",
      "twitter": "https://x.com/jxmnop",
      "talks": [
        {
          "title": "Jack Morris — Context Is Not Memory, Updating Weights Is",
          "description": "A case for when context is enough, and when updating weights may be the real memory mechanism.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06522167,
        0.05576362,
        0.082104616,
        0.06667408,
        0.025116406,
        -0.08318557,
        -0.035366274,
        0.01664424,
        0.08225831,
        -0.2438483,
        0.006547187,
        -0.07495648,
        -0.043509077,
        0.01624406,
        -0.042346347,
        -0.012483411,
        0.074961856,
        -0.039341513,
        -0.048542038,
        -0.040623467,
        0.019647773,
        -0.063296184,
        -0.030461792,
        0.085636355,
        -0.044155773,
        0.062224146,
        0.02368525,
        0.0038257502,
        -0.19600584,
        0.40628937,
        -0.13287084,
        -0.06433231,
        -0.0456189,
        0.003398175,
        0.1588773,
        -0.04426967,
        -0.005202145,
        0.01898807,
        0.056164816,
        0.03801146,
        -0.0055545466,
        -0.010026364,
        -0.03128386,
        0.09479848,
        -0.034840494,
        -0.015886066,
        -0.015996424,
        0.0009889005,
        0.002992333,
        -0.0426545,
        -0.05550666,
        0.019233868,
        0.055190656,
        0.026729103,
        0.06448051,
        -0.0054621217,
        0.03815434,
        -0.035555404,
        0.09512668,
        -0.059710864,
        -0.107541576,
        -0.030758606,
        -0.02804201,
        -0.0023961726,
        -0.0016603997,
        0.0145060485,
        0.035429537,
        -0.07267074,
        -0.018793287,
        -0.022396613,
        0.018817356,
        0.012891127,
        -0.07309147,
        0.055524386,
        -0.10900448,
        -0.0670676,
        -0.082223155,
        0.026765555,
        -0.051547673,
        0.03067973,
        0.033717744,
        0.06430433,
        -0.02470514,
        -0.056119446,
        0.0054896316,
        -0.064233035,
        -0.0050291875,
        0.06798048,
        -0.07056118,
        0.01891055,
        0.089160696,
        0.0019660064,
        -0.09802299,
        0.013552307,
        -0.032943584,
        -0.037992053,
        -0.025345644,
        0.07813657,
        0.101837404,
        0.041138254,
        -0.030474007,
        -0.012805602,
        0.060664803,
        -0.10616916,
        -0.013920154,
        -0.04861511,
        -0.064811565,
        0.00042243506,
        0.06191514,
        -0.0036731411,
        -0.055153333,
        -0.6105714,
        0.004641134,
        0.003982968,
        -0.020244787,
        -0.1446638,
        -0.010053697,
        -0.014832823,
        -0.06130113,
        0.0058886986,
        0.022100491,
        0.0029492236,
        0.021371994,
        -0.0074066776,
        0.0012044421,
        0.110463865,
        0.06511847,
        -0.03478927
      ]
    },
    {
      "id": "worldsfair-speaker-169",
      "conference": "worldsfair",
      "name": "Jonathan Gordon",
      "role": "",
      "company": "ReWeaver AI",
      "twitter": "https://x.com/reweaver_ai",
      "talks": [
        {
          "title": "The Design-Code Roundtrip That Isn't",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.000051471015,
        0.049775112,
        -0.010549636,
        -0.051954757,
        -0.0009478345,
        -0.06536148,
        -0.07028197,
        -0.004931918,
        0.046742044,
        -0.22148237,
        -0.05363469,
        -0.024813782,
        0.040631015,
        -0.040063594,
        -0.028006775,
        0.0009540107,
        0.11966546,
        -0.09376479,
        -0.035901304,
        -0.09517026,
        -0.043740254,
        -0.08127596,
        -0.055973276,
        -0.01724802,
        0.01443343,
        0.0877293,
        0.017500073,
        -0.038099237,
        -0.066149116,
        0.4776768,
        -0.06082652,
        0.034543775,
        0.07293928,
        -0.03652333,
        0.00347858,
        -0.04999776,
        0.015473877,
        0.023405174,
        0.029673597,
        0.063506514,
        -0.048472,
        0.013722717,
        -0.10301398,
        0.017909171,
        -0.044327885,
        0.027781544,
        -0.13931854,
        -0.08117705,
        0.00567008,
        -0.036834847,
        -0.08271396,
        0.037134394,
        0.071704656,
        -0.055659916,
        0.024074612,
        0.0029599925,
        -0.027536316,
        -0.056043036,
        0.07960506,
        -0.034019567,
        -0.04408236,
        -0.014294207,
        -0.022395512,
        0.020112667,
        0.0027464007,
        0.0110787535,
        0.040669456,
        -0.0826785,
        0.022117319,
        -0.04744989,
        0.04875536,
        0.039271645,
        -0.055045113,
        0.057332456,
        -0.08634784,
        0.00054331176,
        0.0009171575,
        0.044407256,
        -0.072961956,
        0.029787676,
        0.08276161,
        0.00077076483,
        -0.09053975,
        -0.0026772758,
        0.017173948,
        -0.042113207,
        -0.07481924,
        0.07490227,
        0.0064071137,
        0.03256612,
        0.019293811,
        -0.031434108,
        -0.0041639856,
        0.03237634,
        -0.06503056,
        -0.008840179,
        -0.0757025,
        -0.0048777866,
        0.06618195,
        -0.053469803,
        -0.08625319,
        -0.00755161,
        -0.014505915,
        0.00090941606,
        -0.12618834,
        -0.09262201,
        -0.01341801,
        -0.013692117,
        0.07470819,
        -0.005665529,
        0.02825099,
        -0.6062081,
        -0.041375253,
        -0.063664,
        -0.009818146,
        -0.1625589,
        -0.011648541,
        0.009353131,
        -0.035270724,
        0.07577696,
        0.017911753,
        0.038484294,
        -0.023583278,
        0.066712186,
        0.0038754966,
        0.057508532,
        0.01468834,
        0.038899284
      ]
    },
    {
      "id": "worldsfair-speaker-170",
      "conference": "worldsfair",
      "name": "Corey Gallon",
      "role": "Managing Director",
      "company": "Rexmore",
      "twitter": "https://x.com/CoreyGallon",
      "talks": [
        {
          "title": "The Dark Arts of Web Automation: Teaching Agents to Use Websites Like Humans",
          "description": "Anything you can do in a browser, your agent can do too. Not by tiptoeing through an MCP server one polite, token-burning call at a time -- properly, programmatically, the way you'd drive any other tool. I'll show you how with chrome-agent, an open source wrapper over the Chrome DevTools Protocol that has become irreplaceable in my everyday work. If you'll ever do a browser task more than once, step-by-step MCP browsing is slow, brittle, and bills you tokens for every single click. A CLI straight onto CDP makes the whole browser programmable: loop it, pipe it, script it, walk away. Write it Tuesday, run it a thousand times Wednesday, all without a second of AI agent babysitting. We'll dispel the MCP hype and myths, with successful demonstrations of cheeky things like: the power of CLI-based browsing and how its so much more capable than mere MCP; reaching through those oh-so-clever cross-origin iframes to clear the verify you're human checkboxes; showing that a JavaScript .click() is not a click, rather, just a function call in a costume that is banhammerable; ultimately, proving that a CDP browser operates just like a meatbag with a mouse and keyboard. You'll learn how to point your AI agents at real, messy, uncooperative websites and web applications and have them get things done exactly the way that you would.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.017521981,
        0.12478938,
        0.018882887,
        -0.020363515,
        0.06828497,
        -0.11321491,
        -0.05408079,
        -0.04510619,
        0.005927401,
        -0.23208553,
        -0.004913357,
        -0.02863288,
        0.049046207,
        -0.008835658,
        -0.011811251,
        -0.016410172,
        0.058326684,
        -0.13917229,
        -0.07350442,
        -0.122121505,
        -0.04183702,
        -0.119232275,
        -0.075816505,
        0.027420001,
        -0.007382018,
        0.11088676,
        -0.051112544,
        0.0035660781,
        -0.11285805,
        0.4347048,
        -0.024198184,
        -0.00557784,
        0.040772725,
        -0.036438454,
        0.06578679,
        -0.010204235,
        -0.015803076,
        0.033016168,
        -0.025408616,
        0.016730236,
        0.014464897,
        0.026112271,
        -0.036995195,
        0.083995566,
        -0.0026940466,
        0.018845184,
        -0.014435669,
        0.0047892104,
        -0.035206314,
        -0.10253342,
        -0.05373454,
        -0.0070870724,
        0.046805836,
        -0.017332625,
        0.0022086725,
        -0.030872576,
        0.022798546,
        -0.025260597,
        0.056115355,
        -0.02421379,
        -0.052748214,
        0.016776126,
        -0.051709007,
        0.03561661,
        -0.042647127,
        -0.062056758,
        -0.008838368,
        0.00012836656,
        0.049443785,
        -0.035772588,
        -0.0060315053,
        0.06599356,
        -0.021550344,
        0.064151175,
        -0.08437482,
        -0.044746697,
        -0.11302142,
        0.053374182,
        -0.0015180114,
        0.009715254,
        0.00444674,
        0.022614341,
        -0.07170477,
        -0.036807075,
        0.01634428,
        0.0034336662,
        -0.058440957,
        0.035207268,
        -0.026745249,
        0.027999552,
        -0.018069567,
        0.0060704225,
        -0.082136646,
        -0.010736566,
        -0.06956764,
        0.01640093,
        -0.06521746,
        -0.025140757,
        -0.005288974,
        0.041161623,
        -0.046446618,
        -0.03094036,
        0.07611761,
        -0.061970536,
        -0.06567689,
        -0.08881618,
        0.02972083,
        -0.009572925,
        0.0528652,
        0.0110684605,
        -0.06927298,
        -0.6479153,
        0.019015765,
        -0.029820053,
        0.0067624804,
        -0.143864,
        0.012390789,
        0.0476823,
        0.011475929,
        0.019367028,
        0.0038093855,
        0.052318364,
        0.00094601256,
        -0.013650124,
        -0.05700123,
        0.06009937,
        0.06873315,
        0.047054242
      ]
    },
    {
      "id": "worldsfair-speaker-171",
      "conference": "worldsfair",
      "name": "Ajay Prakash",
      "role": "",
      "company": "Linkedin",
      "twitter": "https://x.com/ajay_prakash_ai",
      "talks": [
        {
          "title": "500 Skills, Zero Fine-Tuning: LinkedIn's Playbook for AI Agents That Actually Know Your Codebase",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.051865105,
        0.053737342,
        -0.0025029073,
        0.013147574,
        -0.005951829,
        -0.014194051,
        -0.08028307,
        0.033123456,
        0.03710701,
        -0.21556695,
        -0.0511738,
        -0.00084235077,
        0.011629509,
        -0.037306756,
        -0.0037024205,
        -0.040261436,
        0.10751781,
        -0.037069514,
        -0.07079545,
        -0.10345013,
        -0.11753734,
        -0.07555558,
        0.0107660415,
        -0.0044553196,
        -0.050026696,
        0.08088138,
        -0.034034964,
        0.032127276,
        -0.09092175,
        0.4828818,
        -0.10804233,
        0.015463107,
        0.03955946,
        -0.07492084,
        0.033810362,
        -0.030744292,
        -0.050053217,
        0.024358017,
        -0.0066693663,
        0.044283744,
        -0.0080112135,
        -0.013843087,
        -0.052505076,
        0.07493085,
        -0.04335386,
        0.03960664,
        -0.06299218,
        -0.024135116,
        0.0031014066,
        -0.06951391,
        -0.06742978,
        -0.051855844,
        0.010785291,
        -0.06026381,
        0.036863573,
        0.05562701,
        -0.012613487,
        -0.093294024,
        0.075186506,
        -0.063470125,
        -0.05942978,
        -0.016433142,
        0.026352445,
        0.031993054,
        -0.000049035512,
        -0.0649088,
        0.067484446,
        -0.12632127,
        -0.012297504,
        -0.012214405,
        -0.008593518,
        -0.019351467,
        -0.066627525,
        0.06839679,
        -0.088418566,
        -0.075318985,
        -0.033278104,
        0.081102714,
        -0.03746519,
        0.009643183,
        -0.021305343,
        0.06553291,
        -0.06454635,
        -0.04096481,
        0.033870608,
        -0.030844823,
        -0.08985003,
        -0.019794881,
        -0.04247871,
        0.044799384,
        0.022598123,
        0.0013345699,
        -0.00069048745,
        -0.03999437,
        -0.09349855,
        0.03542096,
        -0.06594287,
        0.06273121,
        -0.006464689,
        0.0027066881,
        0.0027832303,
        -0.020468894,
        0.03154876,
        -0.005020878,
        -0.004034525,
        -0.021598328,
        0.013968131,
        -0.043320388,
        -0.015979046,
        -0.08001752,
        -0.049223978,
        -0.6139418,
        0.0067876545,
        0.015559543,
        -0.0065959617,
        -0.1434873,
        0.02488287,
        -0.06279215,
        0.032251075,
        0.09844389,
        -0.002481499,
        0.06530784,
        -0.030361056,
        0.04604136,
        0.035501964,
        0.045565482,
        0.070039265,
        -0.012307015
      ]
    },
    {
      "id": "worldsfair-speaker-172",
      "conference": "worldsfair",
      "name": "Keiji Kanazawa",
      "role": "",
      "company": "Microsoft",
      "twitter": "https://x.com/gojira",
      "talks": [
        {
          "title": "I Let Agents Refactor My Codebase for 3 Weeks. Then I Read the Code.",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.037305582,
        0.021053104,
        0.02127972,
        0.0013705845,
        0.008478734,
        0.030805426,
        -0.052192412,
        0.025193159,
        0.055525996,
        -0.2694798,
        -0.10636399,
        0.029189643,
        0.0061393054,
        0.051366992,
        0.00041250556,
        -0.039083604,
        0.060802553,
        -0.07627075,
        -0.061020218,
        -0.033078324,
        -0.07275133,
        -0.0774387,
        0.011058186,
        0.009933862,
        -0.0028078419,
        0.11206529,
        -0.040911574,
        -0.0028493127,
        -0.056778796,
        0.4759559,
        -0.05805566,
        0.03941671,
        0.04072509,
        -0.057172988,
        0.018007426,
        -0.10392627,
        -0.035210133,
        -0.032143924,
        0.046586733,
        0.04887052,
        -0.09375338,
        -0.03565403,
        -0.051479228,
        0.02963616,
        -0.06657926,
        -0.002616771,
        -0.10691469,
        -0.05699757,
        -0.030320738,
        -0.065199494,
        -0.07343548,
        -0.061935067,
        0.025585132,
        -0.031172385,
        0.027918486,
        -0.0041661803,
        0.004421175,
        -0.009775338,
        0.023220945,
        -0.0064893346,
        -0.010725362,
        -0.001204891,
        0.040490024,
        0.0067175333,
        -0.04445024,
        0.036264647,
        0.053484697,
        -0.05285496,
        0.026158476,
        -0.019285338,
        0.052292235,
        0.0013766238,
        0.0068981177,
        0.010370146,
        -0.1068422,
        -0.053393237,
        -0.017544083,
        0.041773096,
        -0.0015226629,
        0.024517754,
        0.024988452,
        0.08224891,
        -0.07676176,
        -0.065905854,
        0.033641964,
        -0.036473073,
        -0.0863638,
        0.03630131,
        -0.057780664,
        0.04684154,
        0.08682378,
        0.032325447,
        -0.0685081,
        -0.010572535,
        -0.09173647,
        0.011616001,
        -0.038126506,
        0.05710442,
        -0.029610284,
        0.027011933,
        -0.057759784,
        -0.05706564,
        -0.023947328,
        -0.019265844,
        0.0043764585,
        -0.049432952,
        -0.007691859,
        -0.053476803,
        0.08612793,
        -0.026178638,
        -0.01693253,
        -0.6087451,
        -0.09026971,
        -0.041736674,
        0.028258327,
        -0.14653404,
        0.004211174,
        -0.03083375,
        0.0795477,
        0.09236344,
        0.026572768,
        0.08293411,
        -0.011450964,
        0.035831165,
        -0.017027376,
        0.08766116,
        0.035145037,
        -0.032026183
      ]
    },
    {
      "id": "worldsfair-speaker-173",
      "conference": "worldsfair",
      "name": "Ankit Jain",
      "role": "",
      "company": "Aviator",
      "twitter": "https://x.com/ankitxg",
      "talks": [
        {
          "title": "How to Kill the Code Review",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.01856015,
        0.114676744,
        0.019721506,
        -0.03142198,
        0.030516166,
        0.00016918885,
        -0.0890288,
        0.060934648,
        0.005910754,
        -0.18802585,
        -0.05707081,
        -0.0024410796,
        0.021379612,
        -0.019348234,
        -0.0037797773,
        -0.024117233,
        0.079426184,
        -0.098881364,
        -0.04071781,
        -0.04637584,
        -0.09187835,
        -0.06678363,
        -0.03700062,
        -0.011163871,
        -0.019524066,
        0.10160566,
        -0.078047924,
        0.0651712,
        -0.16213812,
        0.46291652,
        -0.03899777,
        -0.025871703,
        0.036399513,
        -0.03203184,
        0.015584674,
        -0.07345173,
        -0.026296206,
        0.009034956,
        -0.016954044,
        0.06037529,
        -0.03839786,
        -0.01361866,
        -0.04566218,
        0.06505552,
        0.005129293,
        0.042219028,
        -0.09502991,
        -0.06906439,
        0.021410797,
        -0.091789536,
        -0.106356576,
        -0.0111813275,
        -0.03602198,
        -0.049481627,
        0.01314654,
        0.076131895,
        -0.07616244,
        -0.052034196,
        0.028438887,
        -0.014862522,
        -0.05204122,
        -0.10265326,
        0.042304493,
        0.014509439,
        -0.02593409,
        -0.07889873,
        0.0953378,
        -0.032140985,
        -0.022838932,
        0.047599904,
        0.04064601,
        0.042420596,
        -0.052747287,
        0.045714103,
        -0.03826222,
        0.024286548,
        -0.006827292,
        0.11465874,
        -0.03406887,
        0.028756674,
        0.067691796,
        -0.003566237,
        -0.09593631,
        -0.03436907,
        0.009639683,
        -0.09943283,
        -0.060009193,
        0.024404122,
        -0.00995069,
        0.07460505,
        0.012873596,
        -0.024983445,
        0.004946555,
        -0.028367717,
        -0.038414165,
        0.08318798,
        -0.020099606,
        0.11225,
        0.05193435,
        0.0067151296,
        0.056535516,
        -0.019562002,
        0.053084735,
        -0.007117372,
        -0.0030660345,
        -0.089856,
        -0.06321263,
        -0.076799646,
        -0.015200676,
        -0.046723746,
        -0.024390994,
        -0.5792319,
        0.060557254,
        -0.0243201,
        -0.06852633,
        -0.14903714,
        0.10141015,
        -0.029287204,
        -0.037273332,
        0.110150225,
        -0.05621567,
        0.0647058,
        -0.048753116,
        0.019006839,
        -0.019550093,
        0.07844662,
        -0.043809757,
        0.0032541356
      ]
    },
    {
      "id": "worldsfair-speaker-174",
      "conference": "worldsfair",
      "name": "Jake Broekhuizen",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Harness Engineering and Continual Learning in LangSmith",
          "description": "Short Description: A practical session on building evaluation and feedback loops that help agents improve over time. We'll look at how teams can use LangSmith to connect traces, datasets, evaluations, human feedback, and production observations into a continual learning loop. Full Abstract: This session will explore harness engineering as a practical way to build better agents. I'll talk about how LangSmith can help teams inspect real behavior, turn traces into datasets, run evals, capture feedback, and use that loop to continually improve their systems. The goal is to give people a simple mental model for moving from ad hoc debugging toward a more systematic way of learning from production and making agents better over time.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06463029,
        0.052181184,
        0.111892655,
        0.0319694,
        0.0042411457,
        -0.052754607,
        -0.07484884,
        -0.008592627,
        0.06823124,
        -0.20827124,
        -0.037578974,
        -0.014402053,
        0.0041332524,
        0.032356348,
        -0.0057995007,
        -0.0063191927,
        0.19144046,
        -0.074132495,
        -0.07028369,
        -0.08726298,
        -0.08024489,
        -0.08896617,
        -0.018707022,
        0.05614652,
        -0.062297802,
        0.12038013,
        0.026176762,
        0.06637171,
        -0.065284245,
        0.42213896,
        -0.07393728,
        0.046219595,
        -0.018554753,
        -0.016262084,
        0.0006445,
        -0.090828724,
        -0.0033665153,
        0.027718658,
        -0.016416026,
        0.124280624,
        -0.01638788,
        0.044757586,
        -0.013986364,
        0.044133693,
        -0.06050733,
        -0.021468194,
        -0.061746642,
        -0.0051974063,
        -0.002242728,
        -0.025408717,
        -0.094357945,
        -0.08474935,
        0.09760797,
        -0.0117838485,
        0.043880615,
        0.055579267,
        0.01684148,
        -0.07099504,
        0.09376416,
        0.019852579,
        -0.053251892,
        -0.01460866,
        0.016249537,
        -0.03683096,
        0.0059033516,
        -0.036788046,
        0.051272366,
        -0.050640304,
        0.010564102,
        -0.03166828,
        0.07688984,
        0.020204708,
        -0.06715879,
        0.015698496,
        -0.060594175,
        -0.122614205,
        -0.06834118,
        0.03998347,
        0.001523343,
        -0.0048862505,
        -0.06250633,
        0.057775278,
        -0.047773752,
        -0.059975293,
        0.009138763,
        0.0131423175,
        -0.048435796,
        0.013293016,
        -0.0729488,
        0.025848635,
        0.04429635,
        0.07093485,
        -0.038954493,
        -0.06517231,
        -0.057553016,
        -0.009610809,
        -0.02203815,
        0.015191703,
        0.022567028,
        0.07241042,
        -0.08154836,
        -0.05033652,
        0.0038587216,
        0.02252861,
        -0.0124067515,
        -0.06807196,
        -0.06217864,
        -0.016508684,
        0.09928722,
        0.047286037,
        -0.026238354,
        -0.6162206,
        -0.027066007,
        0.012119622,
        -0.0535783,
        -0.14660676,
        0.011058806,
        0.070639506,
        -0.00824347,
        0.08974266,
        -0.012704588,
        0.038953103,
        -0.028099217,
        -0.016504139,
        -0.0044588633,
        0.047257554,
        0.051507417,
        0.017645868
      ]
    },
    {
      "id": "worldsfair-speaker-175",
      "conference": "worldsfair",
      "name": "Dan Adler",
      "role": "",
      "company": "Sourcegraph",
      "twitter": "",
      "talks": [
        {
          "title": "The Enterprise Agentic Gap: When Developer-Level AI Tools Hit Millions of Lines",
          "description": "Agentic coding tools have transformed individual developer workflows but owning a large codebase with millions of interdependent lines across multiple code hosts is a different problem entirely. Off-the-shelf AI coding tools weren't built for it, and at scale, they break down in ways that aren't obvious until you're already in trouble. This talk covers the failure modes you'll hit when applying developer-level agentic tools to enterprise-scale migrations, and how Sourcegraph's agentic migrations solution was built to solve what others couldn't.",
          "day": "Day 3 — Session Day 2",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.015731975,
        0.012349244,
        -0.004271996,
        -0.014022735,
        -0.038478225,
        -0.02891374,
        -0.09493667,
        -0.030853791,
        0.05038484,
        -0.22385263,
        -0.057881646,
        -0.056232173,
        0.042447202,
        -0.0046955864,
        -0.047766376,
        -0.017237829,
        0.11391593,
        -0.08973932,
        -0.023830378,
        -0.05214505,
        -0.04253791,
        -0.07706562,
        0.0063369116,
        0.019349268,
        0.019130562,
        0.14540933,
        0.039844524,
        0.08181705,
        -0.10531422,
        0.44840828,
        -0.12750624,
        0.025504252,
        0.0610664,
        -0.04331573,
        0.047347337,
        -0.11877265,
        -0.07267626,
        -0.018144019,
        0.0019707733,
        0.039582208,
        -0.023432685,
        0.019320011,
        -0.08079612,
        0.02749516,
        -0.042691693,
        -0.013491646,
        -0.06791862,
        -0.03267502,
        -0.04247638,
        -0.054774784,
        -0.12251367,
        -0.0023894594,
        0.040017366,
        -0.06438444,
        0.038176242,
        0.03926761,
        -0.009518689,
        -0.02049955,
        0.118719146,
        -0.075684324,
        -0.063370675,
        -0.009535834,
        -0.00066893775,
        -0.008544128,
        0.019379206,
        -0.06360969,
        0.035302676,
        -0.056116726,
        0.05323584,
        -0.00411221,
        0.01731016,
        0.015763279,
        -0.07077513,
        0.025527136,
        -0.060958225,
        -0.110160016,
        -0.0042246063,
        0.062268272,
        -0.058987346,
        0.01599805,
        0.0071989,
        0.046654142,
        -0.1353162,
        -0.032931637,
        0.0719689,
        0.008166383,
        -0.022925911,
        0.008933721,
        -0.013241956,
        0.05303935,
        -0.013507028,
        -0.046372782,
        -0.026146598,
        0.026845807,
        -0.10958377,
        0.020396905,
        -0.05456022,
        0.037386183,
        0.030409064,
        0.018114453,
        0.006373936,
        -0.046785247,
        -0.0030491075,
        -0.031208592,
        -0.011222658,
        -0.070929535,
        -0.022076815,
        -0.025924414,
        0.007675232,
        -0.017136104,
        0.012608985,
        -0.61117685,
        -0.01102978,
        -0.028798485,
        -0.004771276,
        -0.15844464,
        0.05981139,
        0.0008126608,
        0.03440819,
        0.10904686,
        -0.014608807,
        0.02809466,
        -0.035779834,
        0.004310413,
        -0.033860393,
        0.021467607,
        0.101304546,
        0.00016143404
      ]
    },
    {
      "id": "worldsfair-speaker-176",
      "conference": "worldsfair",
      "name": "Elie Bakouch",
      "role": "Researcher",
      "company": "Prime Intellect",
      "twitter": "https://x.com/eliebakouch",
      "talks": [
        {
          "title": "auto-nanogpt",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.021099446,
        0.023862824,
        0.025210774,
        0.02734974,
        0.01071202,
        0.018638391,
        -0.114146486,
        0.053794697,
        0.04883309,
        -0.23073384,
        0.042108446,
        0.0004981062,
        -0.06965291,
        0.019435348,
        -0.058548342,
        0.021089599,
        0.10613971,
        -0.12905672,
        -0.059029188,
        -0.060190808,
        -0.0460698,
        -0.10072908,
        0.007171413,
        0.027717654,
        -0.088547245,
        0.09381079,
        -0.011007598,
        -0.0008848998,
        -0.09533112,
        0.4647964,
        -0.10015795,
        0.016765941,
        0.004462524,
        -0.047168754,
        0.027158124,
        -0.045586396,
        0.003610615,
        0.057596628,
        0.052725878,
        0.071888536,
        -0.048151936,
        0.02974351,
        0.004348477,
        0.024294753,
        0.041856483,
        0.04890806,
        -0.050067604,
        -0.025682643,
        0.00081511063,
        0.037724543,
        -0.016147071,
        -0.0073922146,
        0.03187817,
        0.011434487,
        0.027134972,
        0.03428308,
        -0.03299239,
        -0.06084809,
        0.08841413,
        -0.02254695,
        -0.07620893,
        -0.021418389,
        0.0610244,
        0.049271062,
        0.0289491,
        0.022528669,
        0.004094267,
        -0.07616845,
        -0.042767864,
        -0.031273168,
        0.014085464,
        0.023261247,
        -0.06876022,
        0.00016576074,
        -0.10959349,
        -0.058566175,
        -0.01903869,
        0.11673124,
        -0.04425042,
        0.010412013,
        0.05683072,
        0.09779286,
        0.017451605,
        -0.034270283,
        0.010244761,
        -0.027347805,
        -0.06412331,
        0.05985451,
        -0.038511675,
        0.04239224,
        0.06356255,
        0.0019092005,
        -0.0366191,
        -0.00773466,
        -0.0054515265,
        0.021820279,
        -0.025071952,
        0.09321132,
        0.076841354,
        0.013590212,
        0.020509735,
        -0.049495876,
        0.14262986,
        -0.0275725,
        0.0015762862,
        -0.024578728,
        0.0027581034,
        0.025035752,
        0.10062354,
        0.0005190398,
        -0.024287498,
        -0.6147072,
        -0.028676338,
        0.013323554,
        0.0023843255,
        -0.14954208,
        -0.01448517,
        0.045322634,
        0.013172273,
        0.046739854,
        -0.09686934,
        0.041376047,
        -0.047894664,
        0.013094323,
        0.030051246,
        0.074919015,
        -0.014931898,
        -0.011459256
      ]
    },
    {
      "id": "worldsfair-speaker-177",
      "conference": "worldsfair",
      "name": "Samuel Colvin",
      "role": "",
      "company": "Pydantic",
      "twitter": "https://x.com/samuelcolvin",
      "talks": [
        {
          "title": "Your agent needs a sandbox, not a desert",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.008872578,
        0.032240126,
        0.04729015,
        0.01886733,
        0.013851642,
        -0.05838091,
        -0.046491474,
        -0.059016246,
        0.09081838,
        -0.22199509,
        0.010069142,
        -0.040812865,
        0.014998677,
        -0.07278642,
        -0.0017108724,
        -0.024910511,
        0.08492856,
        -0.086551115,
        -0.026443616,
        -0.050670836,
        -0.08879055,
        -0.10462989,
        0.014695324,
        0.055989973,
        -0.051924355,
        0.08284698,
        -0.06903134,
        0.028310815,
        -0.07705622,
        0.48609027,
        -0.07241732,
        0.04921969,
        0.04102293,
        -0.030600537,
        0.04905065,
        -0.045302186,
        0.04588041,
        -0.012873209,
        0.0033615148,
        0.009514403,
        0.039254688,
        -0.010898082,
        -0.07806546,
        0.037968967,
        0.0044244025,
        0.0107119335,
        -0.0018005366,
        -0.03941329,
        -0.030345153,
        -0.08616999,
        -0.032215834,
        0.023623928,
        0.026995694,
        0.022501217,
        0.053062387,
        0.0139999725,
        -0.042830992,
        -0.09102018,
        0.07729777,
        -0.0032591817,
        -0.07220643,
        -0.047729492,
        -0.004954996,
        -0.004758723,
        0.034809645,
        -0.06487674,
        0.051218685,
        -0.042195648,
        -0.0017615058,
        -0.024096252,
        0.031247739,
        0.027925806,
        -0.0586472,
        0.024960091,
        -0.05267193,
        -0.052206177,
        0.0143581955,
        0.08699644,
        -0.05459992,
        0.046604764,
        0.035123702,
        0.06593645,
        -0.02758012,
        -0.054234535,
        0.0042066067,
        -0.04698195,
        -0.06419253,
        0.034143727,
        -0.0019662406,
        0.010525452,
        0.04588568,
        -0.04084882,
        -0.084011994,
        -0.018966464,
        -0.110020034,
        -0.015461141,
        -0.0046688314,
        -0.030043315,
        0.010119025,
        -0.027196266,
        0.0022070073,
        -0.057964392,
        0.062968396,
        -0.050773967,
        -0.05681281,
        -0.027147539,
        -0.015795019,
        -0.024542203,
        0.056828633,
        -0.06548675,
        -0.051870137,
        -0.62332636,
        -0.07632011,
        -0.063473776,
        -0.040154975,
        -0.14625077,
        0.0656792,
        0.04323673,
        0.016613578,
        0.09989492,
        -0.01392933,
        0.06996215,
        0.040456623,
        0.010448451,
        0.05450644,
        0.060829442,
        0.059590332,
        0.039753594
      ]
    },
    {
      "id": "worldsfair-speaker-178",
      "conference": "worldsfair",
      "name": "Maximillian Piras",
      "role": "",
      "company": "Yutori",
      "twitter": "https://x.com/MVXMXM",
      "talks": [
        {
          "title": "Mousepower: agents that can’t be measured, can’t be managed.",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03711028,
        0.016758645,
        -0.02537968,
        0.041023917,
        -0.018106516,
        -0.053509366,
        -0.08232879,
        0.016135646,
        0.008722923,
        -0.22564456,
        0.00053216115,
        0.02989951,
        -0.0032963953,
        0.006155594,
        -0.038598567,
        -0.008678775,
        0.065952145,
        -0.04636468,
        -0.028169336,
        -0.0740184,
        -0.060631357,
        -0.11749337,
        -0.042705607,
        -0.049112406,
        -0.048635855,
        0.068742305,
        -0.011982105,
        0.014800855,
        -0.15346447,
        0.48491874,
        -0.10575029,
        0.021853112,
        0.025469536,
        -0.058168888,
        -0.019240536,
        -0.0023768246,
        -0.044932805,
        0.029777998,
        0.032473285,
        0.037519045,
        -0.02117937,
        0.013995604,
        -0.06557326,
        -0.031311747,
        -0.079512015,
        0.048116747,
        -0.055613022,
        -0.061046798,
        0.00049093534,
        -0.0740977,
        -0.12396425,
        0.08231395,
        0.0012281187,
        -0.06693485,
        0.02384863,
        -0.017562712,
        -0.048068654,
        -0.028734751,
        0.11210178,
        -0.057706475,
        -0.046645783,
        -0.016817724,
        -0.0039392207,
        0.05158672,
        0.013299556,
        -0.039903004,
        0.052651275,
        -0.07788205,
        0.023603912,
        -0.024215912,
        -0.027095003,
        0.03856538,
        -0.03533186,
        -0.016504202,
        -0.07193377,
        -0.0014838506,
        -0.05736347,
        0.089337476,
        -0.0032643331,
        -0.0126385875,
        0.07826159,
        0.02479007,
        -0.044700194,
        0.0063132183,
        -0.015614186,
        -0.060285017,
        -0.024392864,
        -0.017627686,
        -0.07084043,
        0.040162995,
        0.060165565,
        -0.03240479,
        -0.009082677,
        -0.012563762,
        -0.03840045,
        -0.029062793,
        -0.045403518,
        0.027599499,
        0.036658082,
        0.05115546,
        0.03253657,
        -0.049835343,
        0.09732242,
        -0.03066703,
        -0.015664766,
        -0.062579274,
        0.019463543,
        -0.016007012,
        0.07005483,
        -0.04484366,
        -0.022298452,
        -0.6188929,
        -0.06719491,
        0.020956466,
        0.017515764,
        -0.1615018,
        -0.009554431,
        0.021519711,
        -0.016586507,
        0.07858293,
        0.012772798,
        0.03482703,
        -0.10137351,
        0.051186215,
        0.02376291,
        0.047749687,
        -0.025027541,
        0.007860213
      ]
    },
    {
      "id": "worldsfair-speaker-179",
      "conference": "worldsfair",
      "name": "Paul Klein IV",
      "role": "Founder & CEO",
      "company": "Browserbase",
      "twitter": "https://x.com/pk_iv",
      "talks": [
        {
          "title": "Bringing agents onto the world wide web",
          "description": "The web wasn't built for agents. Heavy HTML, human-first UIs, and a DOM that can hijack the model's context. Still, agents browse it for millions of hours every month through Browserbase, across teams like Ramp, Shopify, and Lovable. This talk walks through that browser agent harness layer by layer, from the security boundary between DOM and model to caching, Agent Identity, and the infrastructure that provisions browsers at scale, and where browser agents go once it is in place.",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.042946544,
        0.1171612,
        0.070979394,
        -0.02366455,
        -0.033287335,
        -0.022046156,
        -0.08520958,
        -0.043124992,
        0.01207197,
        -0.18133168,
        -0.030844256,
        -0.03927668,
        0.04692951,
        -0.055486627,
        -0.07822177,
        0.022459913,
        0.099650435,
        -0.063844584,
        -0.08608002,
        -0.04227511,
        -0.05306762,
        -0.04372012,
        -0.0070327558,
        0.022102783,
        -0.02588977,
        0.039348174,
        0.030065639,
        0.050714567,
        -0.085991055,
        0.48527074,
        -0.06850938,
        0.05787203,
        0.046285834,
        0.042354673,
        0.04555174,
        -0.04739853,
        0.008996075,
        -0.041137718,
        0.03157258,
        0.00316565,
        0.026288334,
        0.05624233,
        -0.026932217,
        0.011029884,
        -0.032733113,
        -0.017527359,
        -0.012215937,
        -0.033745553,
        -0.030245487,
        -0.02394777,
        -0.058515348,
        0.007521533,
        0.017096067,
        0.009490719,
        0.054984465,
        -0.0037006366,
        0.003720086,
        -0.111271106,
        0.006862962,
        -0.036018692,
        -0.07965073,
        0.018437177,
        -0.0040790695,
        0.005433116,
        0.00875201,
        -0.077597514,
        0.017657747,
        -0.03910761,
        0.040471993,
        -0.0542239,
        -0.027740918,
        0.010775949,
        -0.03232755,
        0.025756774,
        -0.11011009,
        -0.10594526,
        -0.0864625,
        0.07790703,
        0.010268934,
        -0.022429826,
        0.005936215,
        0.08286254,
        -0.037968423,
        -0.07645377,
        0.006576184,
        -0.051736128,
        -0.04075199,
        0.07487824,
        -0.05053481,
        0.09046017,
        0.006908619,
        0.004148012,
        -0.054735906,
        -0.027091036,
        -0.07855659,
        -0.009741664,
        -0.014523061,
        -0.013093899,
        0.02842084,
        -0.008117757,
        -0.023006987,
        -0.0133438585,
        0.07053992,
        -0.034480363,
        0.008321811,
        -0.039059777,
        -0.002804998,
        0.023002416,
        0.07175939,
        -0.015855564,
        -0.017490923,
        -0.64564246,
        0.010092946,
        0.035113115,
        -0.04967003,
        -0.15542646,
        0.033850916,
        0.01113239,
        0.014848691,
        0.0046735993,
        -0.07740603,
        0.08256227,
        -0.018821714,
        0.031743675,
        -0.049563468,
        0.07120594,
        0.07447906,
        0.034859598
      ]
    },
    {
      "id": "worldsfair-speaker-180",
      "conference": "worldsfair",
      "name": "Brandon Waselnuk",
      "role": "Founder",
      "company": "Unblocked",
      "twitter": "",
      "talks": [
        {
          "title": "Your agents lack context: Here's how to fix \"You're absolutely right!\"",
          "description": "Every AI coding tool can generate code. Very few can generate the right code for your organization, because they're missing context. They don't know why your team chose Redis over DynamoDB, what the team decided in a Slack thread earlier today about the auth migration, or which architectural patterns your principal engineers actually enforce in review.\n\nThis talk is a practitioner's guide to building a context engine: the reasoning layer that continuously ingests & synthesizes organizational knowledge across disparate sources into unified, queryable understanding.\n\nI'll walk through the problems you actually have to solve — reasoning across systems that don't agree with each other, searching globally before you can reason, maintaining identity-scoped permissions so every user and agent only sees what they should, and personalizing results based on who's asking and what they're working on.\n\nThese are the engineering challenges that make naive RAG fall short, drawn from real lessons building this at scale.",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        },
        {
          "title": "Beyond RAG: See a relational context engine reduce token burn",
          "description": "In this expo talk we'll give you a free context engine simulator, open source tools, and demo how a context engine works. See how modern engineering workflows with agentic loops and goals produce better quality code and reduce token burn. RAG, while useful, leaves context gaps for humans and agents. A context engine fills those gaps by including real-time, relational, personalized, and permission aware techniques to get high-signal context to humans and agents at runtime.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.01232666,
        0.06047734,
        0.053291872,
        0.022177588,
        -0.040664963,
        0.009048924,
        -0.08763806,
        -0.04072582,
        0.011888484,
        -0.27385885,
        -0.06718974,
        -0.045544766,
        0.07654637,
        -0.020828672,
        -0.07130354,
        -0.040984713,
        0.059846975,
        0.0066188206,
        -0.094191715,
        -0.08641256,
        -0.017892787,
        -0.073720895,
        -0.03063931,
        0.046953674,
        -0.019199895,
        0.05823683,
        -0.027249902,
        0.069868125,
        -0.08094077,
        0.4598555,
        -0.06985348,
        0.022490947,
        -0.018389853,
        0.0025460091,
        0.07287284,
        -0.10749558,
        -0.057413746,
        -0.007145088,
        -0.04430607,
        0.042091332,
        -0.03904973,
        0.029056458,
        -0.032447573,
        0.041744865,
        -0.03426758,
        -0.029276416,
        -0.08870362,
        -0.05196305,
        0.0035045198,
        0.015846126,
        -0.10440191,
        0.010868583,
        0.102235414,
        -0.00021422576,
        0.043164045,
        -0.029126514,
        0.022556692,
        -0.08296459,
        0.061824042,
        0.0020331037,
        -0.09638632,
        0.004417629,
        0.020081617,
        -0.0017711542,
        -0.03540405,
        -0.0288987,
        0.08953666,
        -0.004415571,
        -0.043781035,
        -0.02262835,
        0.028151685,
        0.030676952,
        -0.12558876,
        0.060403474,
        -0.061527543,
        -0.12156573,
        -0.08946366,
        0.07311254,
        -0.0711025,
        0.032682754,
        -0.018723397,
        0.044165518,
        -0.056790072,
        -0.07884425,
        0.025005829,
        -0.00025759733,
        -0.09469377,
        0.08081711,
        -0.06506951,
        0.03394818,
        0.009116504,
        -0.0063442755,
        -0.06768335,
        -0.030188905,
        -0.09626922,
        -0.0411272,
        -0.015217703,
        0.035206288,
        0.016184548,
        0.0056475545,
        0.0036308316,
        -0.0066836374,
        0.041421134,
        0.015541811,
        0.015860911,
        -0.025490182,
        0.0011703062,
        -0.040755436,
        0.029579319,
        -0.018494962,
        0.014137358,
        -0.5976096,
        0.048923172,
        -0.02434179,
        -0.009551996,
        -0.1257046,
        0.055669144,
        0.010688564,
        0.06795185,
        0.07005476,
        -0.020033218,
        0.07178452,
        -0.051924754,
        0.017431771,
        -0.0045949863,
        0.01978501,
        0.09236351,
        -0.048209623
      ]
    },
    {
      "id": "worldsfair-speaker-181",
      "conference": "worldsfair",
      "name": "Imad Touil",
      "role": "",
      "company": "QuantumBlack, AI by McKinsey",
      "twitter": "",
      "talks": [
        {
          "title": "AI-Native Organisations runs on Skills: How to Extract, Structure, evaluate and Scale Them",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.11076872,
        -0.032435942,
        0.015260661,
        0.063174784,
        0.03176955,
        -0.008039007,
        -0.042563546,
        0.0387063,
        -0.015522512,
        -0.23573793,
        0.0021227389,
        0.004226806,
        -0.003912095,
        0.0037608685,
        -0.022139162,
        -0.0027979675,
        0.10557835,
        0.0023399952,
        -0.046796355,
        -0.050961807,
        -0.091071434,
        -0.066604,
        0.025224905,
        -0.04110497,
        -0.10075929,
        0.06123117,
        -0.062906325,
        0.01493018,
        -0.05909302,
        0.49567977,
        -0.06279281,
        -0.04478133,
        0.022148082,
        -0.049958035,
        -0.050457038,
        -0.057859436,
        -0.026932925,
        -0.0084244255,
        0.044483427,
        0.070118204,
        -0.03320784,
        0.009903042,
        -0.04163179,
        0.029469475,
        0.028465431,
        -0.031635128,
        -0.03153258,
        -0.037778948,
        0.009242436,
        -0.121866435,
        -0.029119143,
        0.058766305,
        0.021776844,
        -0.034084577,
        0.024116717,
        0.056351777,
        -0.02273431,
        -0.041333944,
        0.09831705,
        -0.09890621,
        -0.04601888,
        0.012958053,
        0.098447755,
        0.058908526,
        0.15150763,
        0.00396341,
        0.020914003,
        -0.069115594,
        -0.008261034,
        -0.021175187,
        -0.01805159,
        -0.038015317,
        -0.025646998,
        0.0022697481,
        -0.07224449,
        0.005710533,
        0.008737019,
        0.12256093,
        -0.082384944,
        -0.0321163,
        0.011675526,
        0.06947805,
        -0.024521424,
        -0.055620644,
        -0.026768329,
        -0.039051376,
        -0.07423891,
        0.05833893,
        -0.0038431478,
        0.04805885,
        -0.004770964,
        0.0409068,
        0.0014614476,
        -0.06517492,
        -0.075290605,
        0.039009687,
        -0.02970183,
        0.034285367,
        0.06914736,
        -0.010264847,
        -0.04428589,
        -0.03311608,
        0.05639422,
        -0.007311203,
        0.043118257,
        -0.10464353,
        -0.0074288207,
        -0.025580922,
        0.08786809,
        -0.0049325973,
        -0.06645342,
        -0.5846211,
        0.082205154,
        -0.03892749,
        -0.02826067,
        -0.18306643,
        -0.049469393,
        -0.04867033,
        -0.007904162,
        0.04712356,
        0.014363746,
        0.03349239,
        -0.004039123,
        -0.0073556034,
        -0.0003026814,
        0.056044452,
        -0.040356994,
        -0.0013176608
      ]
    },
    {
      "id": "worldsfair-speaker-182",
      "conference": "worldsfair",
      "name": "Laurie Voss",
      "role": "Head of Developer Relations",
      "company": "Arize AI",
      "twitter": "https://x.com/seldo",
      "talks": [
        {
          "title": "The Death of the Code Review",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0129248295,
        0.021028223,
        0.016926834,
        -0.002757894,
        0.022307552,
        -0.030842172,
        -0.0925065,
        0.026996074,
        0.061306972,
        -0.19949873,
        0.0061870627,
        -0.049092818,
        -0.0024408274,
        0.022066854,
        -0.08865185,
        0.0334669,
        0.10873998,
        -0.062476117,
        -0.047108367,
        -0.119772844,
        0.010216564,
        -0.016360464,
        -0.020697815,
        -0.01661814,
        -0.08886985,
        0.1304765,
        0.0021794294,
        0.04591703,
        -0.14658768,
        0.49650055,
        -0.080817,
        0.027854659,
        0.016262453,
        -0.023485545,
        0.087373756,
        -0.044063058,
        -0.012896815,
        0.0044575175,
        0.015710548,
        0.07240963,
        0.0016463806,
        -0.0063569807,
        -0.006499853,
        0.023524191,
        -0.060147204,
        -0.0075951745,
        -0.07036907,
        -0.044530343,
        0.024270808,
        0.009785651,
        -0.027560456,
        -0.040069386,
        0.0065900725,
        -0.0427763,
        0.046849694,
        0.082586505,
        -0.043076508,
        -0.10705098,
        -0.006577521,
        -0.049997382,
        -0.15012065,
        -0.023291895,
        -0.00093771034,
        0.04362067,
        0.025304714,
        -0.0627058,
        0.056278914,
        -0.03784213,
        -0.0630299,
        -0.03842978,
        0.025181007,
        0.083834276,
        -0.060298476,
        0.034737542,
        -0.06647914,
        -0.04129629,
        0.015404711,
        0.08231546,
        -0.0707037,
        -0.0066276146,
        0.0007527441,
        0.0068700504,
        -0.13900767,
        -0.06542265,
        -0.011879869,
        -0.044246003,
        -0.061367486,
        0.036561638,
        -0.04151457,
        0.049823433,
        -0.017230395,
        -0.05445195,
        -0.010363785,
        0.036207687,
        -0.049516566,
        0.03382469,
        0.00097029126,
        0.1219009,
        0.025827255,
        -0.012397054,
        0.0001606308,
        -0.051779877,
        0.050200537,
        0.014514659,
        0.0068044597,
        -0.0056923614,
        -0.040091008,
        0.039434407,
        0.0455568,
        -0.05435903,
        -0.04834153,
        -0.57016647,
        0.05203309,
        -0.06546964,
        -0.06627087,
        -0.15470952,
        0.018234171,
        -0.012104617,
        -0.039004203,
        0.124057665,
        -0.014586244,
        0.010653122,
        -0.043381732,
        -0.03031145,
        0.0044492413,
        0.08359539,
        0.08232678,
        -0.00410845
      ]
    },
    {
      "id": "worldsfair-speaker-183",
      "conference": "worldsfair",
      "name": "Rishab Kumar",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "From Stateless to Stateful: Orchestrating Real-Time Voice & Messaging Agents with Twilio and Amazon Bedrock",
          "description": "We have all had that maddening customer service experience: you text a support line about a delayed flight, receive a confirmation, but when you call in a minute later, the voice agent asks, \"How can I help you today?\" completely blind to the SMS you just sent. This is the \"Channel Amnesia\" problem. While businesses are pouring billions into generative AI, most agents are still built on stateless architectures that forget customer context the second a session ends. In this session, we will cure AI amnesia. You will learn how to orchestrate stateful, production-grade AI agents across SMS and Voice using Twilio Agent Connect and Amazon Bedrock. We will dive into why traditional serverless compute fails stateful agents, how to leverage AWS Fargate for isolated, long-lived sessions, and how to configure Bedrock AgentCore over WebSockets to hit sub-50ms streaming voice latency. No slide-ware here expect a live, cross-channel demo and open-source code you can deploy tomorrow.",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05066174,
        0.058792144,
        0.047900394,
        0.04119849,
        0.06710791,
        -0.03738726,
        -0.10929854,
        0.021435138,
        0.05141975,
        -0.30159402,
        -0.03160196,
        -0.05512219,
        0.024099158,
        -0.028379878,
        -0.024857571,
        -0.014579911,
        0.04986726,
        -0.08618959,
        -0.068319954,
        -0.037277997,
        -0.056604803,
        -0.09483135,
        -0.073589064,
        0.062267885,
        -0.06283179,
        0.08850119,
        0.033771668,
        0.0024789763,
        -0.11030804,
        0.4067802,
        -0.07369404,
        -0.026810749,
        0.00785466,
        -0.047822263,
        0.029602878,
        -0.025346529,
        0.0059689544,
        0.01962852,
        0.042207766,
        0.014914831,
        0.013673895,
        0.019953154,
        -0.032255936,
        0.017139876,
        0.053111788,
        -0.007912587,
        -0.0599315,
        0.07468661,
        -0.03814075,
        -0.008304581,
        -0.102222465,
        0.05353102,
        0.030020373,
        -0.0154711,
        0.034029715,
        0.04516724,
        0.05625141,
        -0.069925286,
        0.08713968,
        -0.0019931917,
        -0.05277861,
        -0.061074775,
        0.04364114,
        -0.053940967,
        -0.014898736,
        -0.083083,
        0.0698482,
        -0.0046347305,
        -0.029756656,
        0.00089659076,
        -0.017558595,
        -0.035191223,
        -0.11912657,
        0.036530726,
        -0.036472406,
        -0.1265736,
        -0.04624567,
        0.094527334,
        -0.029248657,
        0.07082733,
        -0.03253371,
        0.08359252,
        -0.103953585,
        0.0010430943,
        0.043110285,
        -0.0214583,
        -0.0006477255,
        -0.0057565854,
        -0.042637985,
        0.15140478,
        0.034830008,
        0.011968165,
        -0.04085731,
        0.025778562,
        -0.04711576,
        -0.012816099,
        -0.041814417,
        0.025800226,
        0.006466626,
        0.010233524,
        -0.0066498346,
        -0.04148544,
        -0.02607885,
        0.031851422,
        -0.012416852,
        0.034983944,
        0.0290386,
        0.010131778,
        0.013133113,
        -0.02016492,
        -0.006395178,
        -0.6244079,
        0.069980726,
        0.011212795,
        -0.015766095,
        -0.14456312,
        0.03932033,
        -0.026189674,
        -0.008388405,
        0.041312594,
        -0.02603101,
        0.09380226,
        -0.027268974,
        -0.02772957,
        0.022356268,
        0.10616557,
        -0.0005485889,
        -0.055939857
      ]
    },
    {
      "id": "worldsfair-speaker-184",
      "conference": "worldsfair",
      "name": "Anant Srivastava",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Prompt, Memory, Weights: The Architecture Decisions Most AI Teams Make by Accident",
          "description": "The interesting engineering in production AI isn't in the model. Your knowledge lives in files, databases, and APIs: docs, runbooks, conversations, code. The model just reads tokens. So the real architectural question is which path that knowledge takes to inference: into the prompt directly, into memory for retrieval on demand, or into the weights through fine-tuning. Most teams treat these as a ladder. Start with prompts, escalate to RAG, eventually fine-tune, as if each step is a more advanced version of the last. The field is converging on a different answer: they solve different problems. The prompt shapes behavior and constraints. Memory grounds the model in current, citable knowledge. Weights harden specialized reasoning and format. They're not substitutes you graduate between; they're complementary, and the failures come from using one to do another's job. Fine-tuning to teach the model facts it should have retrieved is the classic trap: you bake in knowledge that's stale the day it ships, and you still can't cite it. This is an opinionated take on all three: when each is the right call, when each is a trap, and the part most teams never build, the circulation between them. Memory that captures what the agent does becomes the dataset you fine-tune on; fine-tuning changes what's worth retrieving; the loop compounds. Get the three paths right and they stop being a pipeline you climb and start being an architecture that learns.",
          "day": "Day 3 — Session Day 2",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0629204,
        0.0443599,
        0.03481416,
        0.035771605,
        0.09307599,
        0.008175676,
        -0.04957627,
        0.053170964,
        0.029601268,
        -0.23539363,
        0.005440276,
        0.039825756,
        0.040714215,
        -0.062896796,
        0.019030176,
        0.025701612,
        0.063439116,
        -0.059636123,
        -0.02421467,
        -0.066449,
        -0.048172977,
        -0.08243758,
        -0.090187766,
        0.041335616,
        -0.10658355,
        0.11816138,
        0.00058136607,
        0.03306152,
        -0.17267203,
        0.4298221,
        -0.04855659,
        -0.0075629265,
        0.01783661,
        -0.06897169,
        -0.009616884,
        -0.06040639,
        -0.062020347,
        -0.02072944,
        0.027014624,
        0.09593993,
        -0.007441456,
        0.040271934,
        -0.005745467,
        0.008962346,
        -0.021537213,
        -0.030055838,
        -0.1098322,
        -0.039879117,
        -0.0045808847,
        -0.03287505,
        -0.055665113,
        -0.047151573,
        0.07898179,
        -0.023512717,
        -0.0057180584,
        0.09280594,
        -0.0015920706,
        -0.10894404,
        0.06463228,
        -0.06531329,
        -0.053685732,
        -0.04512131,
        0.02225517,
        0.026220892,
        -0.046679977,
        -0.08828695,
        0.056600265,
        -0.045543022,
        -0.016797187,
        -0.016009858,
        -0.03255348,
        0.033134352,
        -0.046908885,
        0.041220788,
        -0.0436815,
        -0.08748502,
        -0.058936797,
        0.082547106,
        -0.0203991,
        0.004661949,
        -0.066420585,
        0.026414368,
        -0.026026374,
        0.013714484,
        0.045711644,
        -0.085896656,
        -0.089081064,
        0.0527046,
        0.043939438,
        0.09214153,
        0.1052896,
        0.010189528,
        -0.074520804,
        -0.00717849,
        -0.0345585,
        0.010009056,
        -0.0053910394,
        0.10514308,
        0.008276756,
        0.025886899,
        -0.00011398534,
        0.029601686,
        0.060823362,
        0.009308197,
        0.008187581,
        -0.05419482,
        -0.010766434,
        -0.04943854,
        0.029696237,
        0.018485151,
        -0.058150172,
        -0.6187457,
        0.013855102,
        0.023897426,
        -0.056202624,
        -0.1428519,
        0.052191794,
        -0.0148733,
        -0.033149254,
        0.040228598,
        -0.025878876,
        0.058030806,
        -0.019473258,
        -0.07361018,
        0.013601244,
        0.021879004,
        0.049207304,
        0.027711978
      ]
    },
    {
      "id": "worldsfair-speaker-185",
      "conference": "worldsfair",
      "name": "Tim Sweeney",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Closing the Loop: An Autonomous AI Research Agent",
          "description": "The holy grail of agentic AI tooling is the autoresearch loop: an agent that can sift through your experiments, create visualizations, propose a hypothesis, launch a training job, read the results, and try again autonomously. In this session, we'll show new autoresearch capabilities built directly into the W&B Models web and iOS apps. We will demo these live using a real-world fine-tuning project, covering everything from launching jobs and reading loss curves to surfacing outlier runs that consume researcher hours and recommending the next steps. Then you'll learn how the eval-driven development loop in W&B Weave makes agents like this trustworthy. You'll see how production traces become benchmarks, and how only the agents that beat the bar make it to production. Join us to learn the same loop we use to improve our own agentic features.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.019594662,
        0.0028851961,
        0.023040796,
        0.06989125,
        -0.005458898,
        -0.043912683,
        -0.05538252,
        -0.029884959,
        0.05850465,
        -0.27548698,
        -0.05062624,
        0.014835628,
        0.022341544,
        -0.03416573,
        -0.0057850797,
        0.053412117,
        0.09982909,
        0.0047263335,
        -0.048660718,
        -0.018985378,
        -0.105352394,
        -0.05710628,
        -0.065746136,
        0.02080146,
        -0.016030218,
        0.14073099,
        0.029278805,
        0.013800333,
        -0.09951377,
        0.4497443,
        -0.0072386814,
        0.060605656,
        -0.0050242855,
        0.010973932,
        0.0009649246,
        -0.03051458,
        0.00954706,
        0.05485994,
        -0.039665397,
        0.048565585,
        -0.029791255,
        0.010544249,
        0.008972988,
        0.051996846,
        -0.033376187,
        -0.050596703,
        -0.051996976,
        -0.06229426,
        -0.069571525,
        -0.022389757,
        -0.026379175,
        0.028831534,
        0.058484085,
        0.012323379,
        0.030263558,
        0.015555907,
        -0.014197396,
        -0.05859935,
        0.070976146,
        -0.02006949,
        -0.17881875,
        -0.062145185,
        -0.023120092,
        0.021570494,
        -0.037771,
        -0.0612319,
        0.016899928,
        -0.024237722,
        -0.040841423,
        -0.07011053,
        0.018121894,
        0.06442853,
        -0.013612797,
        -0.007827903,
        -0.06369711,
        -0.12344026,
        -0.031657074,
        0.054967023,
        -0.0658759,
        -0.010140366,
        -0.010473293,
        0.044806093,
        0.014074233,
        -0.0990122,
        -0.003003385,
        0.0025265396,
        -0.07480678,
        0.0030269192,
        -0.06792028,
        0.03649025,
        0.0032546036,
        -0.0026073775,
        -0.073290996,
        -0.0067084157,
        -0.00019656257,
        -0.022982424,
        0.00074459746,
        0.022856494,
        -0.041732546,
        0.02580022,
        0.0008185287,
        -0.028795756,
        0.022518009,
        -0.04305614,
        -0.035305705,
        -0.025609149,
        0.051129624,
        0.0067997267,
        0.07035743,
        0.0030754218,
        -0.028964438,
        -0.6457064,
        -0.092973165,
        0.005720088,
        -0.0442359,
        -0.14619675,
        0.006533743,
        0.010156868,
        0.075856686,
        -0.011822515,
        -0.022887737,
        0.050347093,
        -0.019742373,
        0.03606679,
        -0.018214548,
        0.0037529366,
        0.03182961,
        0.027879465
      ]
    },
    {
      "id": "worldsfair-speaker-186",
      "conference": "worldsfair",
      "name": "Deepak Pathak",
      "role": "Co-Founder & CEO",
      "company": "Skild AI",
      "twitter": "",
      "talks": [
        {
          "title": "Frontier Robotics Research",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.033812825,
        0.054004338,
        0.05471521,
        0.03346864,
        0.02580466,
        0.006603074,
        -0.024399856,
        -0.014255678,
        0.090098925,
        -0.24978727,
        -0.033925224,
        -0.025555396,
        -0.014919728,
        0.01792328,
        -0.06386159,
        -0.005521908,
        0.026203549,
        -0.0326317,
        -0.026589088,
        -0.055755652,
        -0.066452034,
        -0.05586922,
        -0.02717531,
        0.078043275,
        -0.057147704,
        0.07673236,
        0.01911643,
        0.061456777,
        -0.08577569,
        0.45143285,
        -0.048603635,
        -0.044350717,
        0.036127165,
        0.0027594527,
        0.040053885,
        -0.070017286,
        -0.037603468,
        0.014765553,
        0.0016680085,
        0.050450176,
        -0.06968933,
        -0.005697738,
        -0.05611698,
        0.0073530725,
        0.029119413,
        -0.011485754,
        -0.124761954,
        0.0058227447,
        -0.0055058086,
        -0.03749369,
        -0.06155818,
        -0.0060975505,
        0.040794,
        0.014606555,
        0.003347509,
        0.09226807,
        -0.05667586,
        -0.14966097,
        0.07583089,
        -0.057220332,
        -0.093748145,
        0.03442543,
        0.028791847,
        0.03366939,
        0.015850728,
        -0.07417711,
        0.06114999,
        -0.09004526,
        0.009861031,
        -0.01120429,
        0.11539981,
        0.001699541,
        -0.026488416,
        0.0688015,
        -0.10687467,
        -0.02970432,
        0.031274658,
        0.09539383,
        0.01918921,
        -0.029676298,
        0.025423829,
        0.076571,
        -0.004676072,
        -0.059995417,
        0.00071127835,
        -0.0358815,
        -0.022412386,
        -0.00916001,
        0.0119669,
        0.0922703,
        0.078589335,
        -0.08375735,
        -0.05429414,
        -0.004494226,
        -0.011580185,
        0.007644254,
        -0.05650567,
        0.06185075,
        0.022456689,
        -0.018198412,
        0.006522771,
        0.0066408245,
        0.017184284,
        -0.024887495,
        -0.005500829,
        0.007286891,
        0.075590044,
        -0.017066047,
        0.06402499,
        -0.010791311,
        -0.06624079,
        -0.6174495,
        0.0017956957,
        0.045349848,
        -0.024322767,
        -0.152143,
        -0.0031293808,
        -0.11666107,
        0.024891507,
        0.06220542,
        -0.06748459,
        0.07460787,
        -0.023085121,
        0.05747126,
        0.011552367,
        0.04640923,
        0.0085734725,
        0.0591105
      ]
    },
    {
      "id": "worldsfair-speaker-187",
      "conference": "worldsfair",
      "name": "John Lindquist",
      "role": "Co-founder",
      "company": "egghead.io",
      "twitter": "",
      "talks": [
        {
          "title": "The Agentic Power User's Playbook: Tips and Tricks for Swarm-Style Agentic Development",
          "description": "You opened a fifth agent tab this morning and immediately lost track of which one was doing what. This workshop is the playbook I use daily to run swarms of agents in parallel: the keyboard shortcuts, layout patterns, supervision habits, and fast-model tricks that turn chaos into a control surface. We'll go hands-on: spawning a wall of agents across tiled panes, routing prompts to the right swarm with fast models, switching contexts in milliseconds, recovering when an agent goes off the rails, and building the muscle memory that separates a one-agent-at-a-time user from a true power user. By the end you'll leave with a stocked toolbelt of concrete shortcuts, repeatable patterns, and workspace habits you can drop into your own setup the same day. No cloud, no platform lock-in: every trick runs on the machine in front of you.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        },
        {
          "title": "The Agentic Power User's Playbook: Tips and Tricks for Swarm-Style Agentic Development (continued 2)",
          "description": "You opened a fifth agent tab this morning and immediately lost track of which one was doing what. This workshop is the playbook I use daily to run swarms of agents in parallel: the keyboard shortcuts, layout patterns, supervision habits, and fast-model tricks that turn chaos into a control surface. We'll go hands-on: spawning a wall of agents across tiled panes, routing prompts to the right swarm with fast models, switching contexts in milliseconds, recovering when an agent goes off the rails, and building the muscle memory that separates a one-agent-at-a-time user from a true power user. By the end you'll leave with a stocked toolbelt of concrete shortcuts, repeatable patterns, and workspace habits you can drop into your own setup the same day. No cloud, no platform lock-in: every trick runs on the machine in front of you.",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        },
        {
          "title": "The Agentic Power User's Playbook: Tips and Tricks for Swarm-Style Agentic Development (continued 3)",
          "description": "You opened a fifth agent tab this morning and immediately lost track of which one was doing what. This workshop is the playbook I use daily to run swarms of agents in parallel: the keyboard shortcuts, layout patterns, supervision habits, and fast-model tricks that turn chaos into a control surface. We'll go hands-on: spawning a wall of agents across tiled panes, routing prompts to the right swarm with fast models, switching contexts in milliseconds, recovering when an agent goes off the rails, and building the muscle memory that separates a one-agent-at-a-time user from a true power user. By the end you'll leave with a stocked toolbelt of concrete shortcuts, repeatable patterns, and workspace habits you can drop into your own setup the same day. No cloud, no platform lock-in: every trick runs on the machine in front of you.",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.02691645,
        0.051362623,
        0.050559487,
        -0.009235936,
        -0.038387295,
        -0.065254726,
        -0.03034734,
        -0.008144085,
        0.048012882,
        -0.25523087,
        -0.05896824,
        -0.021091782,
        -0.0011691955,
        -0.046505604,
        -0.04301142,
        0.052249312,
        0.11171622,
        -0.05807016,
        -0.04839722,
        -0.08833756,
        -0.06851277,
        -0.07435119,
        -0.058925357,
        0.0024955194,
        0.0060255285,
        0.11102112,
        -0.009405716,
        0.033383712,
        -0.10626988,
        0.46177438,
        -0.06278447,
        0.018924417,
        0.03844685,
        0.010805703,
        0.012943316,
        -0.047736797,
        0.005234388,
        0.02660496,
        -0.0040160255,
        0.09515385,
        -0.029388316,
        0.03944233,
        -0.04625986,
        0.005593851,
        -0.038958255,
        0.0396279,
        -0.11906509,
        -0.052759163,
        0.0039331703,
        -0.018331693,
        -0.013456125,
        -0.025904953,
        0.029550634,
        -0.040888976,
        0.06876306,
        0.018456582,
        -0.009217901,
        -0.0630418,
        0.0836983,
        0.0069775325,
        -0.06738975,
        -0.0049537467,
        0.040742323,
        -0.0072081215,
        0.0047613466,
        -0.06528194,
        0.047804046,
        -0.08145453,
        0.015914554,
        -0.101873785,
        0.028117929,
        0.08139687,
        -0.055356793,
        0.01585367,
        -0.11056346,
        -0.0842441,
        -0.08167733,
        0.06556095,
        -0.032548163,
        0.043861013,
        0.024049917,
        0.0027363256,
        -0.1019039,
        -0.094104476,
        0.05671249,
        -0.034070786,
        -0.04476531,
        0.017643845,
        -0.04165751,
        0.033184078,
        0.07782708,
        0.02748489,
        -0.07811099,
        0.053465955,
        -0.08232146,
        -0.036809985,
        -0.06766433,
        -0.008954909,
        0.0670086,
        -0.0045524165,
        -0.009638645,
        -0.031763226,
        0.046011195,
        -0.068417266,
        0.042719077,
        -0.101092316,
        0.027798237,
        -0.0731763,
        0.05494866,
        0.0035032968,
        -0.035660055,
        -0.59372866,
        -0.051490325,
        -0.056406487,
        -0.015249095,
        -0.13039692,
        0.058298353,
        -0.016883982,
        0.03884738,
        0.044662226,
        -0.038570553,
        0.049417518,
        -0.045113605,
        0.014932017,
        -0.045332603,
        0.08887802,
        0.045077678,
        0.028019711
      ]
    },
    {
      "id": "worldsfair-speaker-188",
      "conference": "worldsfair",
      "name": "Chris Souza",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Model Whisperers: How Evals and Prompts Shape Agent Behavior",
          "description": "Getting an AI agent to behave the way you want isn’t just about writing better prompts. In real systems, behavior emerges from a loop: prompts->evals->iteration->feedback. Small changes in any part of that loop can completely change outcomes. We saw this while building a seed asset agent - a system that turns messy, real-world advertising creatives (low quality images, cluttered visuals, heavy text overlays) into clean, reusable assets for downstream Gen AI tools. The agent acts like an editor, simplifying visuals, removing unnecessary elements, and isolating core content so that additional context (like text or CTAs) can be added back in a more controlled, brand-safe way. But the real challenge wasn’t just building the agent - it was making it reliable. And prompting alone wasn’t enough. What actually moved the system forward was how we defined success—and how we used evals to reinforce it. Over time, evals stopped being just a way to measure quality. They became part of how the agent learned what “good” looks like. In this talk, we’ll cover: Why prompting alone doesn’t give you stable agent behavior How evals act like feedback signals, not just scorecards How we built evals sets that reflect the real-world Using agent trace logs to understand why things fail (not just that they fail) How to iterate without breaking things you already fixed By the end, you’ll have a set of patterns you can apply to any system dealing with messy/continuously changing data and how to tweak your prompt and evals to accommodate such changes.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.035061862,
        0.048926875,
        0.07970367,
        0.028602304,
        0.0067047537,
        -0.020676365,
        -0.04925129,
        -0.071318395,
        0.040747743,
        -0.25234044,
        0.0015805975,
        -0.0104383845,
        0.016382556,
        -0.020148193,
        -0.019202411,
        0.02770515,
        0.08020471,
        -0.069159515,
        -0.079015896,
        -0.058904886,
        -0.077215336,
        -0.10719678,
        -0.019550383,
        -0.025548778,
        -0.016762054,
        0.14085706,
        0.0027679703,
        0.07051938,
        -0.0996619,
        0.44590938,
        -0.10293021,
        -0.025622563,
        0.03799436,
        -0.018908717,
        -0.019547833,
        -0.0618845,
        -0.05182671,
        -0.017065924,
        -0.008010224,
        0.006857395,
        -0.0052031237,
        0.06074378,
        -0.011696104,
        0.08946355,
        -0.045636065,
        -0.07682786,
        -0.014671557,
        -0.022204263,
        -0.019540884,
        0.0068472703,
        -0.028003223,
        -0.039762452,
        0.00094073685,
        -0.043336533,
        0.018885778,
        -0.0061196107,
        0.0608967,
        -0.038259953,
        0.14520888,
        0.016439132,
        -0.122122176,
        -0.029590698,
        -0.016109573,
        -0.10572506,
        0.039689515,
        -0.062014658,
        0.055436857,
        -0.05531838,
        -0.0062113735,
        -0.060160633,
        -0.012871576,
        0.024780743,
        -0.07614149,
        0.040802024,
        -0.054441754,
        -0.14871171,
        -0.0424588,
        0.014131139,
        0.0029981139,
        0.027364692,
        -0.0081927655,
        0.0899676,
        -0.008829359,
        -0.035423763,
        -0.0009901759,
        -0.014702942,
        -0.057358235,
        0.022308104,
        -0.048058365,
        0.035174686,
        0.039003327,
        -0.0040887264,
        -0.020044822,
        0.008883943,
        -0.080101736,
        0.04422311,
        -0.09459827,
        0.028407637,
        -0.028753258,
        0.008258836,
        -0.019972108,
        -0.041611876,
        0.056464087,
        -0.0076839584,
        0.09241853,
        -0.037052654,
        0.041663475,
        0.04563115,
        0.07500424,
        0.012460743,
        -0.023076236,
        -0.60412985,
        -0.038736925,
        -0.013558363,
        -0.03536832,
        -0.13989243,
        0.10332754,
        0.017325487,
        0.109444745,
        0.03991208,
        -0.039506573,
        0.036034282,
        -0.019495174,
        -0.039479643,
        -0.07718025,
        0.05024306,
        0.028967982,
        0.0015785506
      ]
    },
    {
      "id": "worldsfair-speaker-189",
      "conference": "worldsfair",
      "name": "Preetika Bhateja",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Model Whisperers: How Evals and Prompts Shape Agent Behavior",
          "description": "Getting an AI agent to behave the way you want isn’t just about writing better prompts. In real systems, behavior emerges from a loop: prompts->evals->iteration->feedback. Small changes in any part of that loop can completely change outcomes. We saw this while building a seed asset agent - a system that turns messy, real-world advertising creatives (low quality images, cluttered visuals, heavy text overlays) into clean, reusable assets for downstream Gen AI tools. The agent acts like an editor, simplifying visuals, removing unnecessary elements, and isolating core content so that additional context (like text or CTAs) can be added back in a more controlled, brand-safe way. But the real challenge wasn’t just building the agent - it was making it reliable. And prompting alone wasn’t enough. What actually moved the system forward was how we defined success—and how we used evals to reinforce it. Over time, evals stopped being just a way to measure quality. They became part of how the agent learned what “good” looks like. In this talk, we’ll cover: Why prompting alone doesn’t give you stable agent behavior How evals act like feedback signals, not just scorecards How we built evals sets that reflect the real-world Using agent trace logs to understand why things fail (not just that they fail) How to iterate without breaking things you already fixed By the end, you’ll have a set of patterns you can apply to any system dealing with messy/continuously changing data and how to tweak your prompt and evals to accommodate such changes.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.093507364,
        0.0746066,
        0.12587593,
        0.029134985,
        0.009128653,
        -0.0042906725,
        -0.076330915,
        -0.010172045,
        0.05818597,
        -0.22309501,
        -0.02074955,
        0.0067954566,
        0.003230191,
        -0.06263277,
        -0.06571515,
        0.07070855,
        0.046129964,
        -0.043453585,
        -0.08202754,
        -0.073664196,
        -0.061441485,
        -0.076518476,
        0.010758571,
        -0.016839951,
        -0.01968529,
        0.12643659,
        -0.025903586,
        0.030255333,
        -0.09777863,
        0.47361335,
        -0.1070443,
        -0.0050461125,
        0.07448007,
        -0.02469732,
        -0.027700556,
        -0.07523505,
        -0.016346402,
        -0.030700402,
        0.023012487,
        0.0018619769,
        -0.006647587,
        0.071063794,
        -0.03945857,
        0.106812336,
        -0.064447686,
        -0.06531842,
        -0.022259794,
        0.0336183,
        0.018654777,
        -0.00056226226,
        -0.044487186,
        -0.039564397,
        0.028910086,
        -0.060427014,
        0.048368696,
        0.034544032,
        0.0016667104,
        -0.047185734,
        0.13823113,
        -0.005530108,
        -0.12254968,
        -0.016744763,
        -0.02316999,
        -0.08171722,
        -0.009775906,
        -0.08414079,
        0.051432755,
        -0.046762984,
        -0.047816467,
        -0.04518157,
        0.02128569,
        -0.012536747,
        -0.059913427,
        0.026775338,
        -0.026367499,
        -0.1582036,
        -0.026950616,
        0.074728355,
        0.0051987865,
        0.045027707,
        -0.04242832,
        0.09368185,
        -0.020463305,
        -0.054058492,
        -0.0027883502,
        -0.0050524194,
        -0.08815986,
        0.044460014,
        -0.04063295,
        0.03497131,
        0.079979636,
        -0.009631531,
        -0.07785476,
        0.04705679,
        -0.10322733,
        0.07328621,
        -0.07139005,
        0.052565772,
        -0.005380017,
        0.032534193,
        0.017235916,
        -0.02167058,
        0.0475184,
        0.023080839,
        0.037378777,
        0.022022186,
        -0.018407483,
        0.06942159,
        0.04502881,
        -0.0053029787,
        -0.031737413,
        -0.5557687,
        -0.0034188407,
        -0.007461145,
        -0.0771444,
        -0.14998643,
        0.089962766,
        0.0108332625,
        0.06409438,
        0.05177999,
        -0.018175097,
        0.064475045,
        -0.038962804,
        -0.03213822,
        -0.057337914,
        0.04963872,
        0.06746219,
        0.05735119
      ]
    },
    {
      "id": "worldsfair-speaker-190",
      "conference": "worldsfair",
      "name": "Daniel Bump",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Model Whisperers: How Evals and Prompts Shape Agent Behavior",
          "description": "Getting an AI agent to behave the way you want isn’t just about writing better prompts. In real systems, behavior emerges from a loop: prompts->evals->iteration->feedback. Small changes in any part of that loop can completely change outcomes. We saw this while building a seed asset agent - a system that turns messy, real-world advertising creatives (low quality images, cluttered visuals, heavy text overlays) into clean, reusable assets for downstream Gen AI tools. The agent acts like an editor, simplifying visuals, removing unnecessary elements, and isolating core content so that additional context (like text or CTAs) can be added back in a more controlled, brand-safe way. But the real challenge wasn’t just building the agent - it was making it reliable. And prompting alone wasn’t enough. What actually moved the system forward was how we defined success—and how we used evals to reinforce it. Over time, evals stopped being just a way to measure quality. They became part of how the agent learned what “good” looks like. In this talk, we’ll cover: Why prompting alone doesn’t give you stable agent behavior How evals act like feedback signals, not just scorecards How we built evals sets that reflect the real-world Using agent trace logs to understand why things fail (not just that they fail) How to iterate without breaking things you already fixed By the end, you’ll have a set of patterns you can apply to any system dealing with messy/continuously changing data and how to tweak your prompt and evals to accommodate such changes.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.07868087,
        0.0052781207,
        0.107154034,
        0.0063379123,
        0.0056499112,
        -0.02451902,
        -0.08258835,
        -0.063905455,
        0.06650067,
        -0.23185177,
        -0.0019166119,
        -0.01822973,
        0.008304819,
        -0.0055654407,
        -0.056523886,
        0.04831845,
        0.13057673,
        -0.03214142,
        -0.07916613,
        -0.060622778,
        -0.0880069,
        -0.10192718,
        -0.008925967,
        0.026864842,
        -0.010226358,
        0.12796178,
        0.053895395,
        0.100497305,
        -0.11304348,
        0.40728945,
        -0.09176527,
        0.030430404,
        0.052260526,
        0.002128798,
        0.005194388,
        -0.088050865,
        -0.02394001,
        -0.016818693,
        0.036284093,
        0.009708066,
        -0.020207323,
        0.018743725,
        -0.048248697,
        0.051189896,
        -0.08325216,
        -0.05042536,
        0.015337819,
        -0.0108372485,
        -0.026425159,
        -0.00016914315,
        -0.028341219,
        0.013205735,
        0.035339776,
        -0.01359659,
        0.028420936,
        0.01902317,
        0.05205151,
        -0.013896818,
        0.12901029,
        -0.018889958,
        -0.115598224,
        -0.0143628465,
        -0.08144188,
        -0.11362162,
        0.04847871,
        -0.063990325,
        0.048090477,
        -0.04809558,
        -0.01332724,
        -0.07291226,
        0.026458891,
        0.017231673,
        -0.100683734,
        0.007864048,
        -0.047396284,
        -0.13969547,
        -0.09612418,
        0.061003238,
        -0.05563428,
        0.049149454,
        -0.04759872,
        0.10038673,
        -0.013992837,
        -0.051559668,
        0.018326959,
        -0.053189427,
        0.017698906,
        -0.0068506314,
        -0.007073415,
        0.058780346,
        0.045232523,
        -0.013111763,
        -0.0478583,
        0.04079817,
        -0.0782374,
        0.056259293,
        -0.08678824,
        0.016878938,
        0.020196974,
        0.03599079,
        -0.00006558097,
        -0.019127421,
        0.050782345,
        -0.02209645,
        0.047869425,
        -0.0036786988,
        0.00010881688,
        0.08597403,
        0.06443495,
        -0.030504646,
        -0.0068289735,
        -0.5945299,
        -0.053285275,
        -0.03382248,
        -0.0501136,
        -0.12992455,
        0.1007634,
        0.010418893,
        0.10240599,
        0.033691145,
        -0.025802203,
        0.046512887,
        0.00044441572,
        -0.05315874,
        -0.025389321,
        0.1021201,
        0.0672971,
        0.055412598
      ]
    },
    {
      "id": "worldsfair-speaker-191",
      "conference": "worldsfair",
      "name": "Gaurav Mishra",
      "role": "",
      "company": "Amazon AGI Lab",
      "twitter": "",
      "talks": [
        {
          "title": "From RL to IRL",
          "description": "Today's agents have to operate in a messy reality of flaky connections, ephemeral credentials, and irreversible actions. They need to navigate real software the way humans do: recovering from failures, learning from feedback, and making sound judgment calls. This talk is about the fundamental changes in RL required to make agents ready for IRL. We'll walk through what it takes for training environments to reflect the complexity of the real world, the perception primitives that let an agent see what a user sees, the harness pieces that help it survive contact with real applications, and the failure modes you only discover when you stop scoring and start shipping.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.026360804,
        0.106542334,
        0.047701426,
        0.025774557,
        0.019583505,
        -0.055854827,
        -0.08910612,
        -0.09308271,
        0.0044904444,
        -0.28132334,
        -0.000946931,
        -0.058412943,
        0.057516545,
        -0.039433587,
        -0.032516774,
        -0.040538978,
        0.104235746,
        -0.06491974,
        -0.07247546,
        -0.08268794,
        -0.022361632,
        -0.09255436,
        -0.051076453,
        0.047059465,
        0.023349106,
        0.09064441,
        0.0055037634,
        0.015561764,
        -0.03767939,
        0.4448809,
        -0.09956738,
        -0.049767584,
        0.0074321507,
        -0.046119064,
        0.03155859,
        -0.095727645,
        -0.02891681,
        -0.0015985642,
        0.03811609,
        0.032070663,
        -0.0077911946,
        -0.004452491,
        -0.05021034,
        0.025021449,
        -0.009889907,
        0.0050101867,
        -0.05314421,
        -0.02034142,
        -0.006138815,
        -0.06106836,
        -0.07388802,
        0.07639425,
        0.064973265,
        -0.05998219,
        -0.001601349,
        0.055037178,
        0.0010897559,
        -0.09453679,
        0.09260824,
        0.009537275,
        -0.06742186,
        -0.05339781,
        -0.020714838,
        -0.023226606,
        -0.006393546,
        -0.04414215,
        0.069601715,
        -0.057749685,
        0.024160432,
        0.0052246763,
        0.029706633,
        -0.015310973,
        -0.0069729267,
        0.014119571,
        -0.00858362,
        -0.10994787,
        -0.027913002,
        0.10734007,
        -0.03419885,
        0.020444483,
        -0.00611767,
        0.06332881,
        -0.059599716,
        -0.0016924624,
        0.033652186,
        -0.012050509,
        -0.016589142,
        -0.009856238,
        -0.075991064,
        -0.008640965,
        0.006401571,
        -0.049320836,
        -0.07410654,
        -0.067634776,
        -0.057733856,
        0.011882925,
        -0.069282405,
        0.0869803,
        -0.008592256,
        0.038978174,
        0.021265259,
        0.039078705,
        0.009216712,
        0.028053915,
        -0.0170768,
        -0.009292226,
        0.04018719,
        0.00711905,
        0.047243856,
        -0.038655285,
        0.013022677,
        -0.62527406,
        -0.053973347,
        0.008195672,
        -0.01329555,
        -0.17932959,
        0.079577945,
        0.0063063013,
        0.0002595891,
        0.002051695,
        -0.04434535,
        0.05543162,
        -0.020572213,
        -0.04397279,
        -0.03241557,
        -0.0065005487,
        0.067480385,
        0.032304663
      ]
    },
    {
      "id": "worldsfair-speaker-192",
      "conference": "worldsfair",
      "name": "Yash Patil",
      "role": "Co-founder & CEO",
      "company": "Applied Compute",
      "twitter": "https://x.com/ypatil125",
      "talks": [
        {
          "title": "Building the AC Context Engine",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.014085216,
        0.04366441,
        0.038088784,
        0.07251079,
        0.035177626,
        -0.045279168,
        -0.08148167,
        0.015389935,
        -0.01324946,
        -0.20214473,
        0.012653064,
        -0.064705975,
        -0.0016632826,
        -0.024840897,
        -0.08128199,
        -0.033520967,
        0.012053734,
        -0.02860292,
        -0.04122404,
        -0.08661589,
        -0.079991035,
        -0.045272704,
        -0.06529929,
        0.03959494,
        -0.005933709,
        0.025399474,
        -0.027331797,
        0.0481403,
        -0.13596097,
        0.46203446,
        -0.056795757,
        -0.0866739,
        0.018358855,
        -0.006072799,
        0.04003629,
        0.05296222,
        0.05416394,
        0.022371853,
        0.027070573,
        -0.016668608,
        -0.03709819,
        -0.025933016,
        -0.06856394,
        0.036210548,
        -0.068918504,
        0.042128596,
        -0.0939306,
        -0.04299162,
        -0.0059420373,
        -0.01688966,
        -0.059566587,
        -0.020404622,
        0.008180453,
        0.01408213,
        0.014028527,
        0.028560031,
        -0.046257682,
        -0.11533594,
        0.054620475,
        -0.07874614,
        -0.062209282,
        0.028495843,
        -0.032623284,
        0.010610072,
        0.050901454,
        -0.043830607,
        0.124396555,
        -0.121109195,
        -0.024452742,
        0.0019632143,
        0.0037251038,
        0.030790372,
        -0.043499224,
        0.039565325,
        -0.029730998,
        -0.015366352,
        -0.04106991,
        0.12592645,
        -0.009111309,
        0.059628442,
        -0.004251571,
        0.031548977,
        -0.055672135,
        -0.095745474,
        -0.038701393,
        -0.027648386,
        -0.046291955,
        0.04225087,
        0.00007062717,
        0.06785962,
        0.0014575485,
        -0.056566045,
        -0.08659513,
        -0.048990548,
        -0.03911113,
        0.021618584,
        -0.08259764,
        0.05353506,
        0.03340569,
        0.02260566,
        0.008052295,
        0.009469088,
        0.03406084,
        -0.014107083,
        -0.062417224,
        -0.04494122,
        -0.00791196,
        -0.016606826,
        0.00895681,
        -0.00020629686,
        -0.09847819,
        -0.6325209,
        -0.017174752,
        -0.0019689854,
        -0.07974476,
        -0.15909676,
        -0.032155413,
        -0.048608035,
        0.10994802,
        0.02569298,
        -0.05197528,
        0.04155911,
        -0.03517721,
        0.017976556,
        0.01193376,
        0.019453786,
        0.003635452,
        -0.033657968
      ]
    },
    {
      "id": "worldsfair-speaker-193",
      "conference": "worldsfair",
      "name": "Olive Song",
      "role": "RL Lead",
      "company": "MiniMax",
      "twitter": "",
      "talks": [
        {
          "title": "Agents at Scale: Inside MiniMax's Model and the Infrastructure Behind It",
          "description": "Olive Song (RL Lead, https://www.minimax.io/https://www.minimax.io/) and Dan Fu (VP of Kernels, https://www.together.ai/https://www.together.ai/) dig into the engineering behind one of the most widely used open model families in the agent ecosystem: how MiniMax built the model for agentic workloads, and what it takes to serve it at scale.\n\nOlive on the model side:\n\nThe RL decisions behind long-context reasoning and tool use\n\nWhat training for agentic behavior actually looks like in practice\n\nDan on the infrastructure side:\n\nWhy agentic workloads break inference engines built for chat: prefill-heavy traffic, high cache hit rates, long-context inputs\n\nThe kernel-level optimizations built for MiniMax's workload profile\n\nHow the two teams collaborate on model launches and ongoing performance work",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06328071,
        0.07146659,
        0.020064175,
        0.025191262,
        0.059959803,
        -0.03374351,
        -0.051210795,
        -0.04065347,
        -0.0024713473,
        -0.27085942,
        0.0050467006,
        -0.009666027,
        0.008602031,
        0.03613319,
        -0.062263276,
        0.025903497,
        0.090770125,
        -0.054479793,
        -0.035998106,
        -0.082026616,
        -0.032551836,
        -0.043648887,
        -0.015671331,
        0.03610242,
        0.019923585,
        0.0529334,
        0.033607945,
        0.043117885,
        -0.09856389,
        0.41094077,
        -0.07925827,
        0.0103103,
        -0.007831571,
        -0.0020659554,
        0.070090346,
        -0.04946466,
        0.021651117,
        -0.0065001007,
        -0.0050245873,
        0.04546269,
        0.00272307,
        -0.011712328,
        -0.058453243,
        0.04991915,
        -0.08956788,
        -0.0072996994,
        -0.021587474,
        0.009318413,
        0.020678759,
        -0.04838999,
        -0.02730491,
        0.038835015,
        0.058050495,
        -0.025680803,
        0.021260073,
        -0.0020591149,
        0.008496313,
        -0.11281378,
        0.077437915,
        0.029076926,
        -0.13162225,
        0.078114375,
        0.0057057613,
        0.0010752791,
        0.06621413,
        -0.040070247,
        -0.018775046,
        -0.047545917,
        0.0028919121,
        -0.056544323,
        0.07540145,
        -0.020203106,
        0.02520457,
        -0.008809028,
        -0.15076986,
        -0.15615985,
        0.018952012,
        0.07061182,
        -0.041431114,
        0.037105087,
        -0.009544239,
        0.071385704,
        -0.0938099,
        -0.043238495,
        0.06458681,
        -0.036649454,
        -0.06527389,
        0.024787588,
        -0.04643286,
        0.014784445,
        0.061677635,
        0.058383744,
        -0.027217884,
        0.015377438,
        -0.117406026,
        0.05064615,
        -0.053979527,
        0.07950818,
        -0.039379414,
        0.037499797,
        0.024320912,
        0.042607907,
        0.08324383,
        -0.04206723,
        -0.04658171,
        -0.037490424,
        -0.026451573,
        0.0011985806,
        0.05704738,
        -0.06586023,
        -0.077875674,
        -0.6129712,
        -0.0047866995,
        -0.00079785014,
        -0.094374485,
        -0.15317751,
        0.043834522,
        0.0008668073,
        0.04595216,
        0.08231981,
        -0.0685595,
        0.07361106,
        0.030213354,
        -0.04256011,
        0.04325928,
        -0.008830542,
        0.016911607,
        0.08134964
      ]
    },
    {
      "id": "worldsfair-speaker-194",
      "conference": "worldsfair",
      "name": "Dan Fu",
      "role": "VP of Kernels",
      "company": "Together AI",
      "twitter": "https://x.com/realDanFu",
      "talks": [
        {
          "title": "Agents at Scale: Inside MiniMax's Model and the Infrastructure Behind It",
          "description": "Olive Song (RL Lead, https://www.minimax.io/https://www.minimax.io/) and Dan Fu (VP of Kernels, https://www.together.ai/https://www.together.ai/) dig into the engineering behind one of the most widely used open model families in the agent ecosystem: how MiniMax built the model for agentic workloads, and what it takes to serve it at scale.\n\nOlive on the model side:\n\nThe RL decisions behind long-context reasoning and tool use\n\nWhat training for agentic behavior actually looks like in practice\n\nDan on the infrastructure side:\n\nWhy agentic workloads break inference engines built for chat: prefill-heavy traffic, high cache hit rates, long-context inputs\n\nThe kernel-level optimizations built for MiniMax's workload profile\n\nHow the two teams collaborate on model launches and ongoing performance work",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09308297,
        0.04211194,
        0.009694907,
        0.049542416,
        0.055243634,
        -0.045287214,
        -0.061967883,
        -0.031022074,
        0.00115461,
        -0.24142662,
        -0.034936134,
        -0.06802372,
        -0.028286254,
        0.031500056,
        -0.06049165,
        0.0037285294,
        0.0878365,
        -0.08345114,
        -0.006303293,
        -0.033609565,
        -0.022277806,
        -0.0804846,
        0.0054516112,
        0.0472621,
        0.027380317,
        0.041582204,
        0.050755106,
        0.0764534,
        -0.09009015,
        0.4015249,
        -0.10869861,
        -0.010658439,
        0.008463235,
        -0.028251773,
        0.043069895,
        -0.10109452,
        -0.012346613,
        -0.006887592,
        -0.015131085,
        0.078575805,
        0.021140557,
        -0.00090299966,
        -0.071986,
        0.009796986,
        -0.12343609,
        -0.029591,
        -0.035041373,
        -0.018424615,
        0.015429334,
        -0.04050863,
        -0.05706931,
        0.03782715,
        0.06577357,
        -0.019263087,
        0.015546843,
        0.005699263,
        0.056987766,
        -0.09667527,
        0.08530279,
        0.013622546,
        -0.1720787,
        0.026881984,
        0.035022497,
        0.011122363,
        0.039100446,
        -0.03521799,
        -0.012501557,
        -0.036208235,
        0.021484999,
        -0.054116707,
        0.06537948,
        -0.012287278,
        0.011176413,
        0.000921704,
        -0.104240574,
        -0.1409038,
        0.018381057,
        0.07078019,
        -0.024229446,
        0.02536411,
        -0.018271843,
        0.09494307,
        -0.08041442,
        -0.05640825,
        0.10135035,
        -0.05252162,
        -0.015388173,
        0.008109014,
        -0.023935255,
        0.072634906,
        0.039128494,
        0.06461691,
        -0.06561913,
        0.008106589,
        -0.08839578,
        0.07271323,
        -0.043520126,
        0.08010246,
        -0.0404761,
        0.009809159,
        0.050605573,
        0.04401925,
        0.100074545,
        -0.01912962,
        -0.05125872,
        -0.07721368,
        -0.0076315585,
        -0.013412148,
        0.053095922,
        -0.0035776084,
        -0.050301008,
        -0.6144955,
        -0.066531,
        0.008013825,
        -0.06969318,
        -0.16705242,
        0.02648729,
        -0.02964206,
        0.03848449,
        0.049930375,
        -0.028048782,
        0.063445695,
        0.069517076,
        -0.040437154,
        0.03146901,
        0.0039250124,
        0.01963617,
        0.03147888
      ]
    },
    {
      "id": "worldsfair-speaker-195",
      "conference": "worldsfair",
      "name": "Ben Kus",
      "role": "CTO",
      "company": "Box",
      "twitter": "",
      "talks": [
        {
          "title": "The Half Life of Agent Infrastructure",
          "description": "TBD — talk on search and retrieval, agentic AI, and enterprise AI over unstructured content.",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.042866167,
        0.016915387,
        0.010276725,
        0.045644697,
        0.01662523,
        -0.06585358,
        -0.0637666,
        -0.002491809,
        -0.0043682074,
        -0.23752561,
        -0.0031552487,
        -0.048588153,
        -0.014877513,
        -0.0018118519,
        -0.023726113,
        -0.018286983,
        0.09102665,
        -0.056066643,
        -0.12907653,
        -0.043239452,
        -0.0443605,
        -0.058438234,
        0.017979465,
        0.062188897,
        -0.058427908,
        0.12864506,
        0.020959834,
        0.002967283,
        -0.09080556,
        0.44718292,
        -0.047910422,
        -0.01544686,
        -0.029494975,
        0.050717387,
        0.072484195,
        -0.010940186,
        -0.060165104,
        -0.003918948,
        -0.002704943,
        0.029644458,
        -0.058277916,
        0.09516145,
        -0.00089251634,
        -0.014002354,
        -0.05140878,
        -0.0528859,
        -0.1004976,
        -0.030905101,
        -0.063106164,
        0.048319053,
        -0.08722408,
        0.055240225,
        0.050784897,
        0.0516985,
        0.048710953,
        -0.025437351,
        -0.0025588113,
        -0.067847304,
        0.059129067,
        0.0125064235,
        -0.1605648,
        -0.0008789427,
        -0.01632214,
        0.034170892,
        0.02072977,
        -0.055480357,
        0.046730902,
        -0.042225327,
        0.0009115153,
        -0.005925919,
        0.008978948,
        0.013677825,
        -0.09391189,
        0.03213025,
        -0.10545798,
        -0.106672265,
        -0.042401526,
        0.10795466,
        -0.027307525,
        0.0629501,
        0.02205553,
        0.07872236,
        -0.0816765,
        -0.03531822,
        0.1024136,
        -0.002610114,
        -0.057001006,
        0.0137915285,
        0.008251026,
        0.07544157,
        -0.016968755,
        0.028743183,
        -0.044077955,
        0.03333384,
        -0.10196389,
        0.04896744,
        -0.04533759,
        0.06027013,
        -0.041465614,
        0.07025441,
        -0.042324275,
        -0.04155029,
        0.019112183,
        -0.008796254,
        0.074034914,
        0.016910316,
        -0.01462301,
        -0.0078063807,
        0.09295937,
        0.0027154272,
        -0.012964612,
        -0.6082075,
        -0.016280804,
        -0.010373661,
        -0.022816587,
        -0.15216157,
        0.019112386,
        0.03234816,
        0.05633021,
        0.03408786,
        -0.0141890915,
        0.061861105,
        -0.04044577,
        0.012674362,
        -0.000820763,
        0.07103832,
        0.011871402,
        0.04162663
      ]
    },
    {
      "id": "worldsfair-speaker-196",
      "conference": "worldsfair",
      "name": "Nicholas Arcolano",
      "role": "",
      "company": "Jellyfish",
      "twitter": "https://x.com/arcolano",
      "talks": [
        {
          "title": "Tokenmaxxing is the New \"Lines of Code\"",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.00097176264,
        0.10566628,
        -0.009875706,
        0.017705822,
        -0.02036224,
        -0.06265246,
        -0.10909303,
        0.005667211,
        0.057909068,
        -0.19131875,
        -0.039450273,
        -0.034533124,
        -0.026657673,
        0.05388257,
        -0.036926482,
        -0.0050771045,
        0.08121149,
        -0.16101067,
        -0.019921621,
        -0.072692975,
        -0.07800532,
        -0.062031552,
        -0.018644985,
        -0.009062054,
        -0.046934806,
        0.13085717,
        -0.01114365,
        0.005203479,
        -0.1501851,
        0.47089767,
        -0.080289945,
        -0.0018733063,
        0.034115244,
        -0.03230223,
        0.055973627,
        -0.057429444,
        -0.051685054,
        0.025482988,
        -0.04414543,
        0.011688525,
        -0.05581192,
        -0.060063012,
        -0.11275198,
        0.021510405,
        -0.033560943,
        0.04256582,
        -0.11452758,
        0.011113176,
        0.020087864,
        -0.009995963,
        -0.11458244,
        0.0139343,
        0.029445013,
        -0.0446137,
        0.094701745,
        -0.054049257,
        -0.0004582671,
        0.010655843,
        0.066750795,
        -0.051103443,
        -0.095900394,
        0.0013293739,
        0.018768486,
        0.049665265,
        -0.012393228,
        -0.05735749,
        0.097037144,
        -0.003048555,
        0.00691318,
        -0.05430773,
        -0.038795236,
        0.02568392,
        -0.051421836,
        0.008701513,
        -0.05286431,
        -0.044042226,
        0.023448773,
        0.11812657,
        -0.0298788,
        0.041819017,
        0.05681474,
        -0.017838262,
        -0.070889264,
        -0.034410052,
        0.04740847,
        -0.017148698,
        -0.01483476,
        0.030048901,
        -0.012314425,
        0.058218345,
        0.009527395,
        -0.033648808,
        0.0032927506,
        0.025385989,
        -0.039492097,
        -0.0383301,
        -0.011008969,
        0.061387878,
        0.02482995,
        0.0064085126,
        -0.015542517,
        -0.073530085,
        0.07656358,
        0.007606514,
        -0.059027035,
        -0.040584855,
        0.01180869,
        -0.068909176,
        -0.018429313,
        0.029019352,
        -0.037008,
        -0.5924682,
        0.041939232,
        -0.0182976,
        0.0013633382,
        -0.14723459,
        -0.008972258,
        0.034973957,
        0.05294134,
        0.094610214,
        -0.0067253904,
        0.031582538,
        -0.1056439,
        0.06375334,
        0.07963878,
        0.042512536,
        0.044252314,
        0.01860044
      ]
    },
    {
      "id": "worldsfair-speaker-197",
      "conference": "worldsfair",
      "name": "Vivek Trivedy",
      "role": "Applied Research Lead, LangChain Labs",
      "company": "LangChain",
      "twitter": "https://x.com/Vtrivedy10",
      "talks": [
        {
          "title": "Improving Agents is a Data Mining Problem",
          "description": "Harness Engineering, Post-Training, Continual Learning...these all boil down to the same underlying substrate - Mining Agent Traces 1. I need to run my agents to collect Traces 2. Understand behaviors from Traces at scale 3. Filter data for \"improvement\" 4. Do an improvement step There's a reason why every continual learning platform ends up looking like an observability platform. It's because Traces are the lifeblood of agent improvement. The mechanism that we use to attempt improvement can vary - Harness Eng, SFT, etc. But without understanding the data agents produce, no algorithm will truly build better agents. The holy grail of Agent Improvement is Continual Learning. Consistently mining data and integrating it into the agent definition over infinitely long time horizons. Today, the easiest way to do that is to build an observability platform and constantly point agentic compute to understand the data that agents produce. We'll walk through the current methods of understanding traces at massive scale and choosing how to integrate them to improve agents across your personal agents, team agents, and entire company.",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.02139337,
        0.075700074,
        0.03773632,
        0.070291914,
        0.048952505,
        -0.012672604,
        -0.069116205,
        0.03008008,
        -0.0023948264,
        -0.19692223,
        -0.018694716,
        -0.077497154,
        0.06721839,
        -0.050145227,
        -0.043268584,
        -0.023506608,
        0.11599922,
        -0.021481875,
        -0.046896037,
        -0.0936204,
        -0.05240369,
        -0.0022325115,
        -0.012163538,
        0.069278955,
        -0.02180586,
        0.1484847,
        0.0162174,
        0.06983554,
        -0.07060572,
        0.4548061,
        -0.08684683,
        -0.034298684,
        0.054373875,
        0.0022704694,
        0.02827904,
        -0.08714145,
        0.033450194,
        0.0022202749,
        0.0014667662,
        0.06727208,
        -0.0037733298,
        0.030358866,
        0.0147748655,
        0.026793752,
        -0.01764662,
        -0.0217619,
        -0.058362257,
        0.014994289,
        -0.03653804,
        -0.04206829,
        -0.09217897,
        -0.03428775,
        0.062773496,
        -0.011440799,
        0.06522143,
        0.08524228,
        0.015954424,
        -0.066410914,
        0.06863242,
        -0.010163786,
        -0.10228681,
        0.033855967,
        0.034649074,
        -0.044352625,
        -0.0042669727,
        -0.05463839,
        0.06289981,
        -0.105022155,
        -0.033056714,
        0.014381908,
        -0.025502589,
        -0.02482082,
        -0.057833537,
        -0.012642218,
        -0.08140356,
        -0.088136576,
        -0.10669269,
        0.10261681,
        0.0062352703,
        0.021705931,
        -0.050023906,
        -0.026616449,
        -0.068636805,
        -0.04126357,
        0.04296958,
        0.015218302,
        -0.0038021489,
        0.051412024,
        -0.07825592,
        0.05870397,
        0.0825867,
        0.0037719558,
        -0.048847307,
        -0.050613925,
        -0.084078915,
        0.05956877,
        -0.035011202,
        0.032190356,
        -0.04048092,
        0.0057840664,
        -0.042514015,
        -0.048101287,
        0.07207683,
        -0.023093624,
        0.059512,
        -0.06353208,
        -0.005628868,
        0.0027774044,
        0.04449865,
        -0.05507273,
        -0.12110984,
        -0.6073319,
        0.002318263,
        -0.0028319894,
        -0.012978796,
        -0.16489859,
        0.042524245,
        -0.009066442,
        0.008307154,
        0.04326767,
        -0.054401293,
        0.06585233,
        -0.044692006,
        -0.02344465,
        -0.072216086,
        -0.0014231923,
        0.051925823,
        0.04233623
      ]
    },
    {
      "id": "worldsfair-speaker-198",
      "conference": "worldsfair",
      "name": "Maor Bril",
      "role": "Principal Software Engineer",
      "company": "Character.ai",
      "twitter": "",
      "talks": [
        {
          "title": "Evaling Video Slop",
          "description": "Everyone is shipping video models. Almost no one is evaling them honestly. CLIP score doesn't catch temporal incoherence. Vibes-based human review doesn't scale. And every \"AI judge\" you wire up will quietly drift away from human preference unless you measure the drift. This is a tactical talk on building real multimodal eval, using JudgeJudy (open-sourced at Character.ai) as the working example. You'll leave with: Why video is different from text. Temporal consistency, shot continuity, narrative coherence, and the metrics that actually capture each (clip_temporal, temporal_consistency, and friends). AI judges, the real version. Custom rubrics, when they work, when they hallucinate, when they collapse to a single dimension and pretend they didn't. The calibration loop. Pearson/Spearman correlation against human scores, automated rubric improvement, detecting systematic judge bias before it costs you a release. Pairwise preference models for video. Training a Qwen3-VL backbone with Bradley-Terry loss to score \"is this slop?\" before it ships. Regression gates in CI. How every AgentX release at Character.ai passes through an eval wall before it reaches users. Closing the loop with JudgeJudy. Correlating eval scores against real telemetry (Amplitude, Statsig) and feeding validated gates back into the runtime. If you're shipping any multimodal output and your eval strategy is still \"the team watches some clips on Friday,\" this is the upgrade. github.com/character-ai/judgejudy",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.009617524,
        0.042213157,
        0.034847077,
        0.016310448,
        0.00893773,
        -0.06711408,
        -0.08173846,
        0.020824503,
        0.06165023,
        -0.23357683,
        -0.020146677,
        -0.024149917,
        0.00036967552,
        0.009488362,
        -0.016186824,
        0.013270064,
        0.084714614,
        -0.054905687,
        -0.0704695,
        -0.16372605,
        -0.04911269,
        -0.05405778,
        0.0018564351,
        0.06198109,
        -0.061518207,
        0.09291441,
        0.06404282,
        0.0404404,
        -0.11492563,
        0.4781911,
        -0.17591016,
        -0.051408183,
        -0.0033279979,
        -0.039927784,
        0.05583012,
        -0.057745773,
        -0.04739509,
        0.0012469273,
        -0.01965837,
        -0.008044029,
        0.0021165945,
        -0.002585908,
        -0.015350073,
        0.04144255,
        -0.01016849,
        -0.04675326,
        -0.03788067,
        -0.01322355,
        -0.013007534,
        -0.024159072,
        -0.00674818,
        -0.030733885,
        0.052138995,
        -0.024593575,
        0.04320387,
        0.035717573,
        0.08144461,
        -0.06555463,
        0.12080669,
        0.003286991,
        -0.088841096,
        0.010525353,
        0.046602163,
        0.009892721,
        0.089402996,
        -0.079489835,
        0.054833394,
        -0.06603445,
        -0.054955043,
        -0.0067863706,
        -0.03342896,
        0.042632185,
        -0.029663783,
        0.027959881,
        -0.02149411,
        -0.10946586,
        -0.019573456,
        0.062978975,
        -0.01125247,
        0.0072897244,
        0.008398071,
        0.03541482,
        -0.07506481,
        0.0240823,
        0.019168602,
        -0.024464736,
        -0.11699837,
        0.07986246,
        -0.0071127866,
        0.044687577,
        0.009134457,
        -0.029953182,
        -0.021822011,
        0.021522252,
        -0.048539754,
        0.034378048,
        -0.032026753,
        0.04282633,
        0.0015638189,
        0.0033684499,
        0.00674705,
        0.0032544727,
        0.056372557,
        0.0030257688,
        -0.06729738,
        -0.058907755,
        -0.013928923,
        0.06708826,
        0.06700685,
        -0.045463607,
        -0.061569348,
        -0.59247416,
        0.015148983,
        -0.024017308,
        -0.05461837,
        -0.14632285,
        0.04270752,
        0.00016861527,
        0.006661341,
        0.093514055,
        -0.030778468,
        0.030456657,
        -0.045467794,
        0.03129856,
        -0.049681842,
        0.052787147,
        0.10465733,
        0.012416436
      ]
    },
    {
      "id": "worldsfair-speaker-199",
      "conference": "worldsfair",
      "name": "Hassan El Mghari",
      "role": "",
      "company": "Together AI",
      "twitter": "https://x.com/nutlope",
      "talks": [
        {
          "title": "The Missing Layer: Design Taste in AI Agents // Stop Letting Your Agents Ship Ugly UIs",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.012128207,
        -0.028438443,
        0.053133436,
        0.01658197,
        0.018717559,
        -0.012865669,
        -0.065049484,
        0.02561195,
        0.06446831,
        -0.2096891,
        -0.01063718,
        -0.07206223,
        0.020418273,
        -0.07065444,
        -0.02844296,
        0.029483143,
        0.10603479,
        -0.032838773,
        -0.029063998,
        -0.07751968,
        -0.022168517,
        -0.06670823,
        0.0081507545,
        0.03223058,
        -0.035066467,
        0.08014734,
        0.010130179,
        0.066483915,
        -0.10199677,
        0.46449786,
        -0.09624699,
        0.03065497,
        0.057448614,
        -0.06486597,
        -0.016144412,
        -0.05300853,
        0.008743154,
        0.012674882,
        -0.0014076438,
        0.09760377,
        -0.017075742,
        0.07892831,
        -0.08018076,
        0.060700487,
        -0.041185703,
        0.021837354,
        -0.06288164,
        -0.09686154,
        -0.008632968,
        -0.0655669,
        -0.040478434,
        0.024024252,
        0.01981636,
        -0.10148497,
        0.032237507,
        0.0053222864,
        0.052597865,
        -0.06711533,
        0.03857932,
        0.023237916,
        -0.06605462,
        -0.017009268,
        0.031248039,
        0.04671993,
        0.1077691,
        -0.010629783,
        0.025504183,
        -0.057301413,
        -0.012880277,
        -0.065474756,
        -0.075119,
        0.028675294,
        -0.052158613,
        0.040333018,
        -0.10470808,
        -0.04417429,
        0.0052170344,
        0.05763385,
        0.016577402,
        -0.012731253,
        -0.009423118,
        0.046637464,
        -0.03663625,
        -0.014737816,
        0.057053726,
        -0.07276602,
        -0.016000967,
        0.061254762,
        -0.0068993145,
        0.046735127,
        0.11085045,
        -0.03370738,
        -0.02966835,
        0.005687096,
        -0.09598571,
        0.024846554,
        -0.04783278,
        0.009809864,
        0.0577942,
        -0.043935765,
        -0.056778003,
        -0.037657455,
        0.01648777,
        -0.01762785,
        -0.06417193,
        -0.061647344,
        0.0030793634,
        -0.036312845,
        0.0909626,
        -0.05704926,
        -0.021104285,
        -0.6256813,
        -0.055857286,
        -0.03648985,
        -0.016905885,
        -0.1556696,
        -0.030222343,
        -0.0012195266,
        0.054898605,
        0.08318773,
        0.0396009,
        0.006350472,
        -0.01628493,
        0.03404223,
        0.053716164,
        0.07480201,
        0.03146929,
        -0.002225998
      ]
    },
    {
      "id": "worldsfair-speaker-200",
      "conference": "worldsfair",
      "name": "Francesco Bonacci",
      "role": "Founder & CEO",
      "company": "Cua",
      "twitter": "https://x.com/francedot",
      "talks": [
        {
          "title": "Computer-Use 2.0: Agents Just Got Multi-Cursor",
          "description": "Computer-use agents still inherit a basic desktop limitation: one machine has one foreground app, one hardware cursor, and one active actor. Once you try to run more than one agent per desktop, they start stealing focus from the user and from each other. We built cua-driver around a different model: multiple agents operating real desktop applications in parallel, each with its own synthetic pointer, while the user's cursor and keyboard stay undisturbed. The key move is to stop treating hardware mouse and keyboard events as the primary automation layer. cua-driver goes one layer lower, into the OS plumbing behind accessibility: UI Automation on Windows, AT-SPI on Linux, and AX on macOS. Those APIs address applications and elements directly, so the OS does not require the target window to be frontmost. A click can land on a background window. A keystroke can reach a hidden one. Multiple agents can act at once because none of them is competing for the singleton hardware mouse. I'll walk through the architecture, the API shape, and the platform-specific traps we hit while making it work across Windows, macOS, and Linux. The live demo is three agents operating on one desktop while the user keeps typing uninterrupted. The goal is to make Computer-Use 2.0 feel concrete: what changes in the stack, what becomes possible, and where the approach still leaks, including Wayland, Chromium DOM surfaces, native canvas apps, and fallback input paths.",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09350132,
        0.0527405,
        0.023554698,
        0.016598962,
        0.0066929706,
        -0.10570646,
        -0.04677234,
        -0.0040255603,
        0.052580107,
        -0.24386093,
        -0.02040149,
        0.009291757,
        0.012522998,
        0.0056453114,
        -0.06304878,
        -0.011998501,
        0.09966964,
        -0.029100165,
        0.0019838053,
        -0.037106764,
        -0.1077194,
        -0.13837953,
        -0.031025307,
        0.07974796,
        -0.0724816,
        0.14289467,
        0.07035714,
        0.07202874,
        -0.13012524,
        0.45033768,
        -0.087965995,
        0.023558216,
        0.104770854,
        -0.029667014,
        0.035247643,
        -0.00465246,
        -0.048036866,
        0.023824463,
        0.01916727,
        0.029246224,
        -0.0148301385,
        0.054516807,
        -0.052790508,
        0.0061748843,
        -0.056624644,
        -0.059340693,
        -0.029137285,
        0.00052867504,
        -0.024252063,
        -0.0071428535,
        -0.062452156,
        0.022724234,
        0.052850906,
        0.014269587,
        0.03365959,
        -0.008499282,
        0.08143855,
        -0.040163536,
        0.076729216,
        -0.021117005,
        -0.08287547,
        -0.069274716,
        0.004874352,
        0.068377405,
        -0.038096897,
        -0.07378086,
        0.115232736,
        -0.05373406,
        0.028349655,
        -0.05802354,
        0.05475458,
        0.07930939,
        -0.012113202,
        0.002584117,
        -0.058569267,
        -0.11256298,
        -0.046293236,
        0.05385183,
        -0.016210185,
        0.017707938,
        -0.011100941,
        0.056218114,
        -0.07521771,
        -0.06918769,
        0.031392466,
        -0.03462029,
        -0.054297972,
        0.008820341,
        -0.08626732,
        0.030906541,
        0.050335888,
        -0.043271028,
        -0.0015792069,
        -0.031604066,
        -0.04389158,
        -0.04594679,
        -0.04442081,
        0.025625719,
        0.013254859,
        0.01742923,
        0.016018372,
        -0.058975764,
        -0.024405677,
        0.0005333444,
        0.08433084,
        -0.048641186,
        0.059015986,
        0.0071056783,
        0.11563377,
        0.005393449,
        -0.0071069794,
        -0.59967,
        -0.030668644,
        0.0037763128,
        -0.016265081,
        -0.13367625,
        0.039677817,
        0.030482575,
        0.036706977,
        0.05238198,
        -0.027970277,
        0.065185,
        0.0198749,
        0.010970717,
        0.0050312798,
        0.03428259,
        -0.019210607,
        0.024652623
      ]
    },
    {
      "id": "worldsfair-speaker-201",
      "conference": "worldsfair",
      "name": "Dillon DuPont",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Computer-Use 2.0: Agents Just Got Multi-Cursor",
          "description": "Computer-use agents still inherit a basic desktop limitation: one machine has one foreground app, one hardware cursor, and one active actor. Once you try to run more than one agent per desktop, they start stealing focus from the user and from each other. We built cua-driver around a different model: multiple agents operating real desktop applications in parallel, each with its own synthetic pointer, while the user's cursor and keyboard stay undisturbed. The key move is to stop treating hardware mouse and keyboard events as the primary automation layer. cua-driver goes one layer lower, into the OS plumbing behind accessibility: UI Automation on Windows, AT-SPI on Linux, and AX on macOS. Those APIs address applications and elements directly, so the OS does not require the target window to be frontmost. A click can land on a background window. A keystroke can reach a hidden one. Multiple agents can act at once because none of them is competing for the singleton hardware mouse. I'll walk through the architecture, the API shape, and the platform-specific traps we hit while making it work across Windows, macOS, and Linux. The live demo is three agents operating on one desktop while the user keeps typing uninterrupted. The goal is to make Computer-Use 2.0 feel concrete: what changes in the stack, what becomes possible, and where the approach still leaks, including Wayland, Chromium DOM surfaces, native canvas apps, and fallback input paths.",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.043212697,
        0.030496906,
        0.0155047,
        -0.0149887595,
        -0.0087515265,
        -0.08461998,
        -0.052698586,
        0.012868484,
        0.029387796,
        -0.24328825,
        0.006852671,
        -0.0003168805,
        0.010392062,
        -0.012853319,
        0.006541559,
        0.017219592,
        0.12099216,
        -0.060933564,
        -0.029052805,
        -0.07344679,
        -0.07700775,
        -0.16453671,
        -0.053726833,
        0.07984389,
        -0.044930153,
        0.15412048,
        0.004679742,
        0.1155956,
        -0.06319715,
        0.3861994,
        -0.076645724,
        0.028347993,
        0.05018038,
        0.03779014,
        0.020024873,
        -0.085871816,
        -0.018710023,
        0.015698053,
        0.06756431,
        0.0021213004,
        0.0009693786,
        0.038012214,
        -0.06328937,
        -0.014484236,
        -0.03741769,
        -0.048722107,
        -0.033920407,
        -0.06671987,
        -0.034541376,
        -0.07043163,
        -0.03205657,
        0.051201936,
        0.031828757,
        -0.05588949,
        0.00066970167,
        0.040587854,
        0.052770797,
        -0.06684366,
        0.13502043,
        -0.001777565,
        -0.034070656,
        0.023904081,
        -0.037431262,
        0.04707814,
        -0.031710017,
        -0.033223454,
        0.10726552,
        -0.057762254,
        0.01012925,
        -0.08644967,
        0.02333452,
        0.023406224,
        -0.03160512,
        0.01931125,
        -0.06026937,
        -0.10025639,
        -0.06728856,
        0.007246332,
        -0.022073671,
        0.11308236,
        -0.0041794823,
        0.047793664,
        -0.04301832,
        -0.08024782,
        -0.020577952,
        0.03282414,
        -0.015496084,
        0.07193774,
        -0.07786601,
        0.062429555,
        0.013229541,
        -0.034461822,
        -0.029997407,
        0.009365859,
        -0.0014152885,
        -0.03861386,
        -0.03891756,
        0.029462654,
        -0.014690461,
        0.011471324,
        0.0071498896,
        -0.081095815,
        0.021353941,
        0.023845857,
        0.07386085,
        -0.091626585,
        0.018859014,
        -0.009904103,
        0.02710961,
        -0.035158124,
        0.041385453,
        -0.64414173,
        -0.10412048,
        0.015891062,
        -0.037690055,
        -0.14477347,
        0.031124951,
        0.009244655,
        0.009919664,
        0.00802232,
        -0.013558832,
        0.04875061,
        0.05236154,
        -0.022508493,
        -0.03631951,
        0.02497998,
        0.06515239,
        0.04295483
      ]
    },
    {
      "id": "worldsfair-speaker-202",
      "conference": "worldsfair",
      "name": "Liad Yosef",
      "role": "AI Lead; co-creator of MCP Apps",
      "company": "MCP Apps",
      "twitter": "https://x.com/liadyosef",
      "talks": [
        {
          "title": "MCP Apps - Extending the frontier",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0182997,
        0.0074309097,
        0.009871781,
        0.05025166,
        0.040307198,
        0.004476397,
        -0.059650972,
        -0.01154697,
        0.059146997,
        -0.17778401,
        0.05303913,
        -0.0070801447,
        0.01651972,
        0.009977552,
        -0.03197377,
        0.04662612,
        0.13571781,
        -0.030877978,
        -0.02526271,
        -0.15596454,
        -0.08803539,
        -0.12668943,
        -0.0135948425,
        0.028697832,
        -0.076412275,
        0.049807783,
        -0.025904171,
        0.0091348775,
        -0.1437854,
        0.4534829,
        -0.07516192,
        -0.03406525,
        -0.027761074,
        -0.05221337,
        0.066990376,
        0.011798404,
        -0.040348355,
        0.027985938,
        -0.039371848,
        0.04355231,
        -0.05741432,
        0.007859171,
        -0.046814114,
        0.008291345,
        -0.02772149,
        -0.0434632,
        -0.08724276,
        0.019659884,
        0.01639171,
        -0.017020684,
        -0.025646571,
        -0.041281734,
        0.061398942,
        -0.013185466,
        0.020321134,
        0.025866693,
        0.04045924,
        -0.005956366,
        0.064690895,
        -0.06482831,
        -0.0447903,
        0.03263363,
        -0.006853883,
        0.068192564,
        0.011786173,
        -0.045075558,
        0.061214432,
        -0.06113045,
        -0.024450773,
        -0.023204096,
        0.012525726,
        0.04865604,
        -0.0037798819,
        -0.008692061,
        -0.13877064,
        0.029163253,
        -0.01994744,
        0.098352596,
        0.012212596,
        -0.008942826,
        0.045338754,
        0.027727338,
        -0.075305484,
        -0.008485747,
        -0.04476043,
        -0.028998366,
        -0.07561729,
        0.103652604,
        0.014073904,
        0.06666199,
        0.045137227,
        -0.023014879,
        -0.023748664,
        0.02447414,
        -0.09209455,
        0.0075886776,
        -0.07408102,
        0.08543603,
        -0.0061650914,
        -0.030234385,
        -0.009343138,
        -0.065273166,
        0.04280277,
        -0.017973892,
        -0.053415153,
        -0.086560965,
        -0.04642722,
        0.02577373,
        0.12209875,
        -0.008366491,
        -0.09839856,
        -0.59683627,
        0.089046076,
        -0.051266607,
        -0.069013186,
        -0.17187268,
        0.015351647,
        -0.044435594,
        0.061172992,
        0.084509574,
        -0.07061878,
        0.07005991,
        -0.05072114,
        0.07282393,
        0.015318122,
        0.04145746,
        -0.025782788,
        -0.03885311
      ]
    },
    {
      "id": "worldsfair-speaker-203",
      "conference": "worldsfair",
      "name": "Ido Salomon",
      "role": "",
      "company": "MCP Apps",
      "twitter": "https://x.com/idosal1",
      "talks": [
        {
          "title": "MCP Apps - Extending the frontier",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0368798,
        0.06171144,
        0.004640345,
        0.0057517113,
        -0.011334217,
        0.0017383553,
        -0.102085836,
        -0.0038293353,
        0.05372743,
        -0.20931034,
        0.016785454,
        -0.036364593,
        0.072175466,
        0.0020780747,
        -0.05180139,
        -0.008678992,
        0.15230861,
        -0.024980692,
        -0.018946128,
        -0.13866393,
        -0.0569394,
        -0.12835528,
        0.00093894737,
        0.014999478,
        -0.035993643,
        0.09918295,
        -0.014495394,
        0.007524175,
        -0.12477916,
        0.46369368,
        -0.051879182,
        -0.013784998,
        -0.057898447,
        -0.061324667,
        0.12736364,
        0.0063585006,
        -0.0005066315,
        0.06870139,
        -0.009862021,
        0.070173964,
        -0.049327873,
        0.014427652,
        -0.09019653,
        0.019747699,
        0.05330771,
        -0.05703737,
        -0.036799703,
        -0.018510237,
        -0.033963013,
        -0.05149888,
        -0.067301385,
        -0.024396446,
        0.0759087,
        -0.007915112,
        0.007498506,
        -0.017120339,
        -0.014974745,
        0.05617265,
        0.056947105,
        -0.0893843,
        -0.028376758,
        -0.009197,
        0.0012220173,
        0.04334103,
        -0.058790687,
        0.017801713,
        0.06701284,
        -0.039829724,
        -0.007827085,
        -0.005054005,
        0.055770822,
        0.07152335,
        -0.009385275,
        0.04062079,
        -0.11071013,
        0.034636978,
        -0.015275228,
        0.035985567,
        -0.012142065,
        -0.028538657,
        0.039060958,
        -0.01543892,
        -0.09586659,
        -0.005648821,
        -0.060181227,
        -0.02300128,
        -0.10796474,
        0.072470196,
        0.03633815,
        0.013711667,
        0.03208512,
        -0.008906962,
        0.022263024,
        0.069333784,
        -0.08042612,
        -0.008285684,
        -0.047243696,
        0.0326427,
        -0.019724745,
        -0.022334585,
        -0.020966914,
        -0.08644603,
        -0.03984035,
        -0.007182452,
        -0.008581024,
        -0.06753828,
        -0.047197837,
        -0.010339633,
        0.09922129,
        0.017969783,
        -0.0229336,
        -0.5799313,
        0.09030224,
        -0.031173147,
        -0.029129263,
        -0.14775428,
        0.0029782907,
        -0.018669505,
        0.114550315,
        0.11073482,
        -0.023666432,
        0.12671743,
        0.014874117,
        0.08468052,
        0.030603703,
        0.0041494514,
        -0.04155199,
        -0.029171657
      ]
    },
    {
      "id": "worldsfair-speaker-204",
      "conference": "worldsfair",
      "name": "Will Brown",
      "role": "Researcher",
      "company": "Prime Intellect",
      "twitter": "https://x.com/willccbb",
      "talks": [
        {
          "title": "PRIME-RL: Async & Decentralized RL Training at Scale",
          "description": "Will Brown (Researcher at Prime Intellect) covers post-training for LLM agents: multi-turn reasoning, credit assignment, distributed RL, PRIME-RL, and verifier-driven environments for LLM RL.",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.011569561,
        0.077177584,
        0.054549012,
        0.010587751,
        0.03861567,
        -0.034480087,
        -0.08088224,
        -0.008893687,
        -0.029069979,
        -0.24045615,
        0.0019536999,
        -0.0823445,
        -0.012930969,
        0.015788086,
        -0.066528365,
        -0.0033238977,
        0.071007445,
        -0.046919875,
        -0.10458182,
        -0.0464025,
        -0.020649903,
        -0.09440735,
        -0.007772814,
        0.118785866,
        -0.03878194,
        0.07337076,
        0.009846236,
        0.058689766,
        -0.11494774,
        0.397498,
        -0.06954916,
        -0.020014107,
        -0.018740552,
        0.03166773,
        0.061592102,
        -0.07712663,
        0.017128117,
        -0.028102065,
        0.07372446,
        0.056834668,
        -0.009267439,
        0.01563894,
        -0.026445407,
        0.01690514,
        0.030038955,
        -0.028607043,
        -0.0022886964,
        -0.01006418,
        0.0056775524,
        0.0055956235,
        -0.09719152,
        0.029061612,
        0.048127208,
        -0.009551441,
        0.043182105,
        0.040991865,
        -0.041133296,
        -0.100414775,
        0.064796604,
        0.017006522,
        -0.066491306,
        0.018006593,
        0.053745057,
        -0.027733786,
        0.047129713,
        -0.043753598,
        0.033175442,
        -0.057700735,
        0.021390842,
        -0.00848044,
        -0.009001666,
        0.0051417835,
        -0.048988633,
        -0.0018918489,
        -0.04379965,
        -0.15434247,
        -0.101167224,
        0.15556106,
        -0.032957032,
        0.06019231,
        0.0054076235,
        0.07458956,
        -0.02112669,
        -0.039417744,
        0.11538617,
        -0.030594083,
        -0.07300503,
        0.02271569,
        -0.08543473,
        0.065863155,
        0.0743213,
        -0.015940027,
        -0.10784339,
        -0.07152601,
        -0.08150666,
        0.06126433,
        -0.017029922,
        0.07085066,
        0.094457835,
        0.056509193,
        0.018081328,
        0.0018296284,
        0.06390087,
        -0.027794447,
        0.006661721,
        -0.027868992,
        -0.0032423292,
        0.045437034,
        0.06682472,
        -0.09141118,
        -0.05636569,
        -0.62076753,
        -0.047792744,
        -0.035164043,
        -0.041337587,
        -0.14787146,
        0.070249416,
        0.012631646,
        -0.013199651,
        0.021242395,
        -0.039124083,
        0.049800377,
        -0.001968269,
        0.043807764,
        -0.0007116619,
        0.0038582715,
        0.016359935,
        0.03007155
      ]
    },
    {
      "id": "worldsfair-speaker-205",
      "conference": "worldsfair",
      "name": "Rachna Srivastava",
      "role": "",
      "company": "DFPI",
      "twitter": "",
      "talks": [
        {
          "title": "Guardians of the State: How We Built an Air-Gapped AI Fortress for Consumer Data",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03592874,
        0.055313252,
        0.01298519,
        0.07113349,
        0.11165737,
        -0.0073101036,
        -0.046235986,
        -0.02055837,
        0.05759828,
        -0.17649993,
        -0.07580431,
        0.008496671,
        -0.08591946,
        -0.008533787,
        -0.015391938,
        0.024686925,
        0.053108845,
        -0.04765929,
        -0.044414632,
        -0.037370026,
        -0.0391989,
        -0.04627427,
        0.047263302,
        -0.0026330594,
        -0.07294053,
        0.08854863,
        0.010014565,
        0.01471676,
        -0.08681303,
        0.50175476,
        -0.08999376,
        0.01754605,
        0.04552078,
        -0.039649002,
        0.10873004,
        -0.0077288668,
        -0.03249545,
        -0.045272976,
        0.043559182,
        0.08509112,
        0.013311486,
        -0.00002912065,
        -0.026585232,
        0.042601094,
        0.00008629406,
        -0.0105970595,
        -0.059852906,
        -0.06254282,
        -0.043768995,
        -0.007100011,
        -0.055345092,
        -0.024377648,
        0.045300547,
        -0.038631815,
        -0.007021023,
        0.0559609,
        -0.026382532,
        -0.11336612,
        0.060257915,
        -0.090919085,
        -0.13308407,
        0.02428945,
        0.006175163,
        0.013341283,
        0.020043366,
        -0.06034589,
        0.065824166,
        -0.0882298,
        0.002537591,
        0.0013815967,
        0.012132674,
        -0.0040454296,
        -0.0537244,
        0.041942686,
        -0.037656292,
        -0.0932614,
        -0.04013818,
        0.056553103,
        -0.018344877,
        0.026466722,
        0.035583023,
        0.07039981,
        -0.08651351,
        -0.043233205,
        0.03006463,
        -0.020842029,
        -0.023338888,
        0.026585903,
        -0.0047180643,
        0.067729756,
        0.024588179,
        -0.0155586945,
        -0.010580064,
        -0.006087796,
        -0.06498505,
        0.041408632,
        -0.06883205,
        0.0061313496,
        0.033653833,
        0.024740456,
        0.008730046,
        -0.05638744,
        -0.062003784,
        0.037473667,
        0.036361452,
        -0.010074814,
        0.0024310176,
        -0.02313886,
        0.051348753,
        -0.012097039,
        -0.07342218,
        -0.6101405,
        0.036938712,
        -0.0030973873,
        -0.028771728,
        -0.15789996,
        -0.008226696,
        -0.039189592,
        0.032397177,
        0.13663463,
        -0.030501263,
        0.07391932,
        -0.10061736,
        -0.047536332,
        0.0660924,
        0.025547683,
        0.018130075,
        0.011721367
      ]
    },
    {
      "id": "worldsfair-speaker-206",
      "conference": "worldsfair",
      "name": "Matthew Jewkes",
      "role": "",
      "company": "Standard Cybernetics",
      "twitter": "https://x.com/mjewkes",
      "talks": [
        {
          "title": "Superhuman performance is a shape, not just nines.",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0072502154,
        0.03083992,
        0.06732614,
        0.07847632,
        -0.007070819,
        0.014552738,
        -0.06373863,
        -0.024784692,
        0.074878156,
        -0.26418456,
        -0.07775869,
        0.01394111,
        -0.0047293147,
        0.022646,
        -0.053104777,
        -0.008260486,
        0.07766243,
        -0.058180764,
        -0.06893083,
        -0.061348174,
        -0.015237286,
        -0.091279775,
        -0.0002269459,
        0.028776957,
        -0.04477939,
        0.10367732,
        0.012883816,
        0.06351249,
        -0.11789656,
        0.49722794,
        -0.052886147,
        -0.008130624,
        0.0065210266,
        -0.040889207,
        0.09330204,
        -0.061061915,
        -0.0072975243,
        0.04982852,
        0.006393041,
        0.06423712,
        -0.00040108553,
        -0.021794992,
        -0.018851742,
        0.043438446,
        -0.0052474015,
        0.035139024,
        -0.038582705,
        -0.061246667,
        0.06253742,
        -0.0395396,
        -0.045977943,
        0.0590607,
        0.03781751,
        -0.013410895,
        -0.01493403,
        0.022678679,
        -0.025621016,
        -0.022062989,
        0.056182116,
        -0.117542624,
        -0.12534454,
        0.044596452,
        0.002472765,
        0.010986894,
        0.024855902,
        -0.06885139,
        0.11329296,
        0.049173366,
        0.015804633,
        -0.011834371,
        0.0074811596,
        0.018112838,
        -0.077345006,
        -0.03921948,
        -0.067416675,
        -0.043634564,
        -0.047489785,
        0.04651668,
        -0.058544077,
        0.055638075,
        0.027196003,
        0.049371462,
        -0.026581006,
        -0.038558196,
        -0.06769264,
        -0.011858636,
        -0.02772438,
        0.05673307,
        -0.027187169,
        0.0034407554,
        0.05228424,
        -0.011515463,
        -0.0055131284,
        -0.010403608,
        0.031467564,
        -0.0021428717,
        -0.07112285,
        0.045837924,
        -0.003561546,
        -0.0047253044,
        -0.044435263,
        -0.06790029,
        -0.0069406186,
        -0.039410222,
        -0.04913216,
        -0.06129818,
        -0.0054009086,
        -0.008500398,
        0.08921506,
        -0.045438364,
        0.02989674,
        -0.60596657,
        -0.0017894497,
        0.003050754,
        -0.015073253,
        -0.14773645,
        0.00082222244,
        0.02280999,
        -0.048055455,
        0.05395831,
        0.010007799,
        0.07309313,
        -0.03775421,
        0.0882529,
        -0.023179475,
        0.0376643,
        -0.0107326815,
        -0.03401692
      ]
    },
    {
      "id": "worldsfair-speaker-207",
      "conference": "worldsfair",
      "name": "Kevin Orellana",
      "role": "Software Engineer",
      "company": "Amazon Web Services",
      "twitter": "https://x.com/KevssOrellana",
      "talks": [
        {
          "title": "1,000 Agent Tasks in a Sandbox: What Breaks When LLMs Write and Run Code",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.033189174,
        0.087131165,
        0.004332639,
        0.005154973,
        0.06960364,
        -0.0591227,
        -0.06257519,
        -0.012268656,
        0.026611242,
        -0.21605252,
        -0.0031712034,
        -0.09072829,
        0.008714945,
        -0.00664481,
        -0.03592866,
        0.011014011,
        0.040539727,
        -0.12882128,
        -0.010204905,
        -0.08171496,
        -0.049281396,
        -0.1066372,
        -0.03601626,
        -0.014404539,
        -0.08827499,
        0.08643054,
        -0.014899341,
        0.06672812,
        -0.07546974,
        0.5057707,
        -0.07342735,
        0.013345351,
        0.05746858,
        -0.019775104,
        0.05456289,
        -0.08792719,
        0.038169533,
        0.012409655,
        0.015718851,
        -0.02359398,
        -0.035421647,
        -0.0002090302,
        -0.08630198,
        0.06276486,
        -0.024522709,
        -0.011578427,
        -0.033880092,
        -0.02631825,
        -0.053902425,
        -0.060021732,
        -0.08292829,
        0.052695096,
        0.065203756,
        0.028161213,
        0.031359717,
        0.014477331,
        -0.03856733,
        -0.0886579,
        0.05553489,
        0.016869176,
        -0.016809812,
        -0.028345946,
        0.012020644,
        0.005619277,
        0.0096181445,
        -0.07413738,
        0.06739754,
        -0.0046867034,
        -0.017614473,
        -0.033885308,
        0.049659695,
        -0.03366257,
        -0.06151456,
        0.0059951902,
        -0.05199323,
        -0.07204159,
        -0.03515413,
        0.10270293,
        -0.020271208,
        0.049647227,
        0.079118654,
        0.08110254,
        -0.08873302,
        -0.001184271,
        0.048810802,
        -0.0011695946,
        -0.05787825,
        0.052242514,
        -0.10191207,
        0.08655804,
        0.015440458,
        0.027284073,
        -0.058325004,
        -0.033994883,
        -0.062734336,
        -0.003768292,
        0.011930909,
        0.035399158,
        0.023983115,
        0.0098411795,
        -0.01916235,
        -0.013909226,
        0.03753151,
        -0.05903299,
        -0.027434576,
        -0.039082855,
        -0.022669073,
        0.0023312974,
        0.07203525,
        -0.045380734,
        -0.00476369,
        -0.6061618,
        0.0005944242,
        -0.10494968,
        -0.03993127,
        -0.13815849,
        -0.02567697,
        0.001252842,
        0.030202314,
        -0.008972632,
        -0.013434857,
        0.07881841,
        -0.032023028,
        -0.02182055,
        0.0136843445,
        0.00881636,
        0.034715973,
        -0.009112488
      ]
    },
    {
      "id": "worldsfair-speaker-208",
      "conference": "worldsfair",
      "name": "Juraj Kabzan",
      "role": "",
      "company": "Skydio",
      "twitter": "",
      "talks": [
        {
          "title": "From Manual Drones to Autonomous Multi-Agent Missions",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.054108866,
        -0.07525018,
        0.052696597,
        0.0687005,
        -0.018847052,
        0.0003991663,
        -0.031065028,
        -0.035799272,
        0.07423442,
        -0.23536626,
        -0.042060986,
        0.013820596,
        -0.011872456,
        0.043388575,
        -0.06479885,
        -0.0034179753,
        0.1436451,
        -0.047648642,
        -0.026184566,
        -0.13192905,
        0.008118964,
        -0.096980095,
        -0.036967427,
        -0.0023736423,
        0.0004420056,
        0.0731375,
        0.0065829214,
        0.037209056,
        -0.06957396,
        0.51191473,
        -0.103183955,
        -0.014097726,
        0.011012448,
        -0.03019433,
        0.045411278,
        -0.088900015,
        -0.07805892,
        -0.013223364,
        0.074914716,
        0.04906816,
        -0.027184846,
        -0.03295431,
        -0.053445682,
        0.02586347,
        -0.048380613,
        0.055588946,
        -0.061387748,
        0.009496126,
        -0.041634317,
        0.008004044,
        -0.0567429,
        0.04812699,
        0.01385958,
        -0.023561021,
        0.034971736,
        0.026094481,
        -0.056239475,
        -0.053722307,
        0.102367595,
        -0.054187328,
        -0.07316664,
        -0.047993015,
        -0.02529785,
        0.047013666,
        0.014487077,
        -0.04358295,
        0.050758637,
        -0.10304139,
        0.0012401304,
        -0.034791395,
        0.0136152,
        -0.020517888,
        0.0049987314,
        0.0047795395,
        -0.08080603,
        0.04249145,
        0.024341194,
        0.08616991,
        -0.031813562,
        -0.028703464,
        0.01988765,
        0.055707175,
        -0.05791608,
        -0.051624805,
        0.04053556,
        -0.008471376,
        0.0015228111,
        -0.01549367,
        0.0077919518,
        -0.0029606537,
        0.08677336,
        -0.007513536,
        -0.029573329,
        0.023386503,
        0.007818901,
        -0.0059004803,
        -0.0569204,
        0.0759521,
        0.020607516,
        -0.033112463,
        -0.033341855,
        -0.0048575336,
        -0.012958218,
        -0.015783187,
        0.035618994,
        -0.027674707,
        0.062310398,
        -0.0076851486,
        0.016767848,
        -0.048319455,
        0.052270066,
        -0.5977504,
        -0.082259834,
        0.016892869,
        -0.030124862,
        -0.16624817,
        0.020772992,
        -0.011226268,
        0.0045154374,
        -0.016345685,
        -0.0898347,
        0.0616597,
        -0.02306465,
        0.08253903,
        0.0568657,
        0.013037603,
        -0.012810841,
        -0.004612891
      ]
    },
    {
      "id": "worldsfair-speaker-209",
      "conference": "worldsfair",
      "name": "Yu Su",
      "role": "Associate Professor, Computer Science and Engineering",
      "company": "Ohio State University",
      "twitter": "https://x.com/ysu_nlp",
      "talks": [
        {
          "title": "From RAG to Memory: Non-Parametric Continual Learning for LLMs",
          "description": "Talk on continual learning for LLMs and agents, drawing on retrieval-to-memory and environment-adaptation research.",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.047868103,
        0.04004329,
        0.020237328,
        0.06483244,
        0.0706459,
        -0.027466284,
        -0.08955105,
        -0.019456923,
        -0.031273592,
        -0.20390168,
        0.00015413144,
        -0.09677951,
        0.025843857,
        0.016188435,
        -0.04802376,
        -0.022736086,
        0.036118284,
        -0.024599923,
        -0.0561555,
        -0.0524614,
        0.062050954,
        -0.03414582,
        -0.01205993,
        0.091755286,
        -0.05198415,
        0.115653515,
        0.013317676,
        0.044728987,
        -0.06868587,
        0.39199644,
        -0.086798266,
        -0.046184983,
        -0.0074972147,
        -0.005321072,
        0.053795982,
        -0.06624036,
        0.012377656,
        -0.066277795,
        0.019973086,
        0.111961864,
        -0.008549208,
        0.0039750226,
        -0.07064607,
        0.064011805,
        -0.046069313,
        -0.091087,
        -0.05177834,
        -0.030270005,
        -0.0514684,
        0.0497493,
        -0.027718686,
        -0.05378728,
        0.050700273,
        -0.010804258,
        0.009734467,
        0.013375153,
        0.049937647,
        -0.10043196,
        0.13313822,
        0.032160368,
        -0.12678294,
        0.0067851255,
        0.018732475,
        0.012536462,
        0.0016873687,
        -0.068003885,
        0.037391797,
        -0.07401745,
        -0.059126597,
        -0.0039531826,
        0.0027037007,
        0.0032175458,
        -0.03244319,
        0.009982235,
        -0.08302273,
        -0.12877159,
        -0.044518366,
        0.045115802,
        -0.061995417,
        -0.0016091845,
        0.041883823,
        0.06834951,
        -0.056773543,
        -0.05019691,
        0.08056863,
        -0.0073752296,
        -0.09694096,
        0.089267805,
        -0.09588377,
        0.07027397,
        0.0336292,
        -0.027264781,
        -0.098114334,
        -0.042390514,
        -0.102673575,
        -0.0035548364,
        -0.039675172,
        0.13450499,
        0.014174741,
        0.059646934,
        -0.03794659,
        0.0051959977,
        0.032858264,
        -0.019063724,
        -0.0053943032,
        -0.057096813,
        -0.0137737105,
        0.008559348,
        0.0724844,
        0.009562855,
        -0.035683244,
        -0.6279624,
        -0.02879526,
        0.018768307,
        -0.050142128,
        -0.15137206,
        0.025986237,
        -0.04086774,
        -0.008809174,
        0.06320134,
        -0.06708056,
        0.044682223,
        0.03001672,
        -0.042386055,
        0.0128883235,
        0.061361443,
        0.0869593,
        -0.008839592
      ]
    },
    {
      "id": "worldsfair-speaker-210",
      "conference": "worldsfair",
      "name": "Akele Reed",
      "role": "",
      "company": "Sondermind",
      "twitter": "",
      "talks": [
        {
          "title": "Evals Driven-Development: Engineering a Mental Health AI Coach Ethically & Safely",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.045804627,
        0.056765754,
        0.07328274,
        0.0057900944,
        -0.028110161,
        0.030463882,
        -0.0070821675,
        -0.056345664,
        0.06464481,
        -0.22919695,
        -0.062478393,
        -0.04862498,
        0.024618546,
        -0.030598136,
        -0.009263391,
        0.061919726,
        0.06287543,
        -0.0329487,
        -0.027012747,
        -0.07017482,
        -0.031857267,
        -0.051504802,
        0.025701335,
        0.019969843,
        -0.074219674,
        0.100222714,
        -0.043710638,
        0.07678393,
        -0.084645465,
        0.48431087,
        -0.12885986,
        0.0021858104,
        0.025284147,
        -0.0062564807,
        0.068534754,
        -0.081364185,
        -0.01360281,
        0.020964177,
        0.025009653,
        0.036888734,
        -0.04572439,
        -0.05390181,
        -0.028469034,
        0.02001197,
        -0.07154896,
        -0.02271801,
        -0.034042858,
        -0.045910563,
        -0.007559206,
        -0.022706617,
        -0.052222766,
        -0.06968903,
        0.04179669,
        -0.014622043,
        0.040267296,
        0.07195894,
        -0.0067261136,
        -0.10332618,
        0.056594,
        -0.025035402,
        -0.047562763,
        -0.024570549,
        0.036919698,
        0.029665263,
        0.0036422233,
        -0.06812238,
        0.040594205,
        -0.073127106,
        -0.028902492,
        -0.08229457,
        0.018521857,
        -0.007811078,
        -0.044620883,
        0.03615798,
        -0.10734012,
        -0.037914712,
        -0.02155256,
        0.10267878,
        -0.04690148,
        0.031182516,
        0.016673068,
        0.02124255,
        -0.09307704,
        -0.05338491,
        0.0014422057,
        -0.05056598,
        -0.07091128,
        0.04180262,
        -0.007881682,
        0.054279074,
        0.009692786,
        -0.007961901,
        -0.055024236,
        -0.0038345626,
        -0.06613187,
        0.02147725,
        -0.10130598,
        0.036979936,
        -0.043192644,
        0.00015417152,
        -0.07163873,
        -0.0333054,
        0.09622905,
        -0.054081324,
        -0.0073986766,
        -0.008456959,
        -0.088696815,
        0.023785222,
        0.09301863,
        0.012898093,
        0.008021477,
        -0.5917663,
        -0.01238839,
        -0.0108630955,
        -0.04572048,
        -0.15167262,
        0.005947245,
        0.027804159,
        -0.0011558746,
        0.14888412,
        -0.04334406,
        0.03098526,
        -0.035716165,
        0.029757714,
        0.05173504,
        0.05375481,
        0.10446392,
        -0.019912979
      ]
    },
    {
      "id": "worldsfair-speaker-211",
      "conference": "worldsfair",
      "name": "Dave Revere",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Evals Driven-Development: Engineering a Mental Health AI Coach Ethically & Safely",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.051600985,
        0.011339591,
        0.042773135,
        0.028725956,
        0.014638399,
        0.06847209,
        -0.064671196,
        -0.012886491,
        0.037020292,
        -0.2104986,
        -0.04631368,
        -0.04070103,
        0.014630006,
        -0.032526992,
        -0.048219956,
        0.03903557,
        0.09673361,
        -0.06691409,
        -0.067850605,
        -0.043821905,
        -0.0646064,
        -0.047020663,
        0.008479146,
        0.0420513,
        -0.01750152,
        0.10651081,
        -0.020835586,
        0.082187615,
        -0.10199653,
        0.48938158,
        -0.12127318,
        -0.018316722,
        0.08128488,
        0.027365003,
        0.08282936,
        -0.04038626,
        -0.047202464,
        -0.01782615,
        0.018190928,
        0.084908634,
        0.019067042,
        0.027400518,
        -0.007652904,
        -0.0038884287,
        -0.06396842,
        -0.015386059,
        -0.041079253,
        -0.08129631,
        -0.033309028,
        -0.026662946,
        -0.06863958,
        -0.046481736,
        -0.012965827,
        -0.065085456,
        0.050439253,
        0.030106025,
        -0.031028293,
        -0.105893046,
        0.04626941,
        -0.06661483,
        -0.12081058,
        0.008463201,
        0.0007045776,
        -0.026847301,
        -0.020563964,
        -0.040128816,
        0.05434091,
        -0.05419873,
        -0.036206674,
        -0.07815463,
        -0.0058872025,
        -0.005956896,
        -0.050878536,
        -0.012163498,
        -0.07190613,
        -0.00917693,
        -0.080252506,
        0.12613592,
        -0.033223562,
        0.010347729,
        0.018728556,
        0.09457347,
        -0.10796525,
        -0.04683604,
        0.008077143,
        -0.024374532,
        0.0064925705,
        0.040718194,
        0.020346157,
        0.061376,
        -0.021917012,
        -0.038470834,
        -0.058771946,
        0.013216936,
        -0.08551582,
        -0.019476105,
        -0.014673598,
        0.055921927,
        -0.039437193,
        0.03940969,
        -0.038673636,
        0.0141689805,
        0.02726778,
        -0.008040184,
        -0.034780502,
        0.007303309,
        -0.035805326,
        -0.009587821,
        0.10187944,
        0.07395776,
        -0.02442318,
        -0.5950112,
        -0.019599421,
        -0.04029163,
        -0.025180312,
        -0.16079499,
        -0.0380293,
        0.042252284,
        0.01041265,
        0.09543745,
        -0.0068816706,
        0.053697594,
        -0.018532235,
        -0.014913293,
        0.041723885,
        0.05785503,
        0.06964628,
        -0.038853284
      ]
    },
    {
      "id": "worldsfair-speaker-212",
      "conference": "worldsfair",
      "name": "Doug Keller",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Evals Driven-Development: Engineering a Mental Health AI Coach Ethically & Safely",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.0076982877,
        -0.020087503,
        0.061873388,
        0.071959905,
        0.035133455,
        0.009703464,
        -0.035321247,
        -0.018669661,
        0.08950667,
        -0.20606156,
        -0.068421975,
        -0.07641678,
        -0.02986771,
        -0.027228793,
        -0.044787437,
        0.07668229,
        0.13295856,
        -0.05399646,
        -0.11323312,
        -0.078618206,
        -0.039043188,
        -0.07658126,
        0.02249153,
        0.044493515,
        -0.020961242,
        0.10743691,
        -0.00398831,
        0.10372252,
        -0.015071275,
        0.48262867,
        -0.14574137,
        0.03693916,
        0.00049855385,
        -0.012476384,
        0.048311803,
        -0.057825513,
        -0.00756738,
        -0.04580405,
        0.029717915,
        0.03535662,
        0.06194493,
        0.0056786276,
        -0.0631145,
        -0.025092628,
        -0.050423823,
        -0.058750343,
        -0.009252571,
        -0.059820835,
        0.0049703466,
        -0.0765886,
        -0.06590045,
        -0.036599085,
        -0.012067516,
        -0.00024282395,
        -0.0023585027,
        -0.00889226,
        0.0022280156,
        -0.06358195,
        0.09264691,
        0.005295911,
        -0.05921427,
        0.03203311,
        0.0033137554,
        0.009582236,
        0.017378058,
        0.011016269,
        0.014199702,
        -0.09479172,
        -0.0383333,
        -0.087239414,
        0.0079316255,
        -0.05217903,
        -0.06860775,
        0.05296289,
        -0.06832849,
        -0.034751374,
        -0.050211985,
        0.10570009,
        -0.1080237,
        -0.02703021,
        0.012242869,
        0.08549369,
        -0.055977892,
        -0.07845071,
        0.0078007635,
        -0.038373653,
        0.035259616,
        0.035027936,
        -0.01195253,
        0.052649375,
        0.0013332675,
        -0.044031646,
        -0.034369677,
        -0.006588804,
        -0.016410178,
        0.016643388,
        -0.04987891,
        0.07777282,
        -0.07132777,
        0.013303335,
        -0.06331661,
        -0.031181479,
        0.02632507,
        -0.040132523,
        0.016732128,
        -0.015122724,
        -0.026397672,
        -0.039154317,
        0.09469094,
        0.0045976434,
        -0.017481102,
        -0.602096,
        0.010960571,
        -0.040270384,
        -0.00690589,
        -0.15074591,
        0.046639074,
        0.020084612,
        0.03732801,
        0.09969058,
        0.0011253698,
        0.012620938,
        0.016083213,
        -0.0042198156,
        0.040856954,
        0.049094103,
        0.066630624,
        -0.019732123
      ]
    },
    {
      "id": "worldsfair-speaker-213",
      "conference": "worldsfair",
      "name": "Dustin Mihalik",
      "role": "Software Engineer, AI Platforms",
      "company": "Indeed",
      "twitter": "",
      "talks": [
        {
          "title": "MCP Apps: Give the Model Data, Give the User a UI",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.025351737,
        0.0003238076,
        -0.053055782,
        -0.002799101,
        0.07037376,
        -0.00583612,
        -0.12388135,
        -0.007894458,
        0.055499755,
        -0.24226937,
        0.021831905,
        -0.017886678,
        -0.0033822907,
        -0.038453486,
        0.026102653,
        0.04026045,
        0.11135956,
        -0.059729755,
        -0.061737888,
        -0.099472314,
        -0.018592302,
        -0.09309289,
        -0.053326424,
        0.031670682,
        -0.060374048,
        0.115439825,
        0.049101748,
        0.11064422,
        -0.044093985,
        0.4216698,
        -0.18154357,
        0.02067074,
        0.022285152,
        0.0011908644,
        0.06112213,
        -0.04576675,
        -0.07068085,
        -0.002350955,
        -0.04305988,
        0.08447191,
        -0.11892003,
        0.0114874635,
        -0.043469056,
        -0.01041661,
        -0.007541097,
        0.013912907,
        -0.088204294,
        -0.017471178,
        0.03614351,
        -0.03236363,
        -0.023592586,
        0.05194368,
        0.008283008,
        -0.03257401,
        -0.036733575,
        0.07045304,
        -0.034698427,
        -0.08439172,
        0.028141856,
        -0.060845777,
        -0.058750458,
        0.06719042,
        0.050981168,
        0.022750992,
        -0.0016157195,
        -0.067968905,
        0.098881505,
        0.0054812324,
        -0.009214979,
        -0.030707566,
        0.11956714,
        0.029633678,
        -0.048894662,
        0.057068337,
        -0.060441293,
        -0.0050710593,
        -0.027770652,
        0.09912346,
        -0.07248256,
        0.06287671,
        0.06335458,
        0.02653723,
        -0.054688476,
        -0.058697056,
        -0.012858365,
        -0.031222966,
        -0.038963653,
        0.0018343832,
        -0.0103354305,
        0.1238269,
        -0.03490054,
        -0.03950391,
        -0.0132583305,
        -0.02561365,
        0.024778483,
        -0.014573344,
        -0.03157961,
        0.064494945,
        0.01419417,
        0.058723506,
        -0.067752294,
        -0.05211982,
        -0.0066455193,
        0.04657491,
        -0.050662525,
        -0.016405938,
        -0.0024851703,
        -0.031685017,
        0.055674862,
        -0.022644354,
        -0.002162992,
        -0.5912521,
        -0.011003042,
        0.009196999,
        0.03442263,
        -0.16613542,
        -0.016140208,
        -0.015454103,
        0.13360925,
        0.05218275,
        -0.0065409555,
        0.035166826,
        -0.07006886,
        -0.02529694,
        0.04267081,
        0.04486865,
        0.103388764,
        -0.018942572
      ]
    },
    {
      "id": "worldsfair-speaker-214",
      "conference": "worldsfair",
      "name": "Kenny Workman",
      "role": "CEO",
      "company": "LatchBio",
      "twitter": "",
      "talks": [
        {
          "title": "LatchBio",
          "description": "Hold for LatchBio. AI-powered biotech platform for biological data infrastructure and multi-omics analysis; user requested inclusion among new AI startups.",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.09659101,
        0.09325492,
        0.016795797,
        0.005589673,
        0.0002767902,
        -0.003541366,
        -0.09001349,
        0.024998989,
        0.017368866,
        -0.23781466,
        0.033912618,
        -0.041220527,
        -0.008106859,
        0.054090142,
        -0.090689585,
        -0.0078666555,
        0.100496516,
        -0.030497616,
        -0.054657992,
        -0.08193089,
        -0.058740832,
        -0.027382277,
        -0.014866879,
        0.039812364,
        -0.07341827,
        0.04527778,
        0.0672411,
        0.009819115,
        -0.06909737,
        0.42745623,
        -0.034889217,
        -0.00043449784,
        0.012105897,
        0.037771035,
        0.038843986,
        -0.02006873,
        -0.028958715,
        -0.0033490274,
        0.01061575,
        0.06032113,
        -0.029978523,
        0.010499214,
        0.07467267,
        0.015257145,
        -0.040816307,
        0.007627812,
        -0.06731799,
        -0.03802505,
        -0.021401864,
        0.024571935,
        0.02705075,
        0.062709324,
        0.026320942,
        0.13218085,
        0.08072391,
        0.04438239,
        -0.0026637553,
        -0.100074075,
        0.03454174,
        -0.057738215,
        -0.075453654,
        -0.04303087,
        0.01546943,
        0.045173343,
        -0.0020100698,
        -0.048113346,
        0.061073408,
        -0.029143779,
        -0.007293747,
        -0.037999574,
        0.06525827,
        0.086654976,
        -0.06558863,
        -0.04630517,
        -0.012649273,
        -0.104368806,
        -0.046739705,
        0.08488346,
        0.038252503,
        0.061297305,
        0.0059956242,
        0.03661114,
        -0.05874436,
        -0.08285505,
        -0.004056633,
        -0.06884947,
        -0.05085668,
        0.01664902,
        -0.023929002,
        0.052268665,
        0.024476593,
        -0.06625099,
        0.02604691,
        -0.033341378,
        -0.0410389,
        0.032666944,
        -0.0073193116,
        0.025735907,
        -0.06536654,
        0.010189973,
        0.011060545,
        -0.019794384,
        0.018665735,
        0.012929944,
        -0.016047997,
        -0.07847259,
        0.025199119,
        0.021152608,
        0.0630588,
        -0.012621573,
        -0.015541914,
        -0.6570218,
        -0.008924587,
        0.02018564,
        -0.047921084,
        -0.13836068,
        -0.01620025,
        -0.060707018,
        0.0817206,
        0.073361754,
        0.019451493,
        0.053123057,
        -0.0055259154,
        0.055327706,
        0.04000132,
        0.044927932,
        0.037847564,
        -0.10817733
      ]
    },
    {
      "id": "worldsfair-speaker-215",
      "conference": "worldsfair",
      "name": "Sunita Verma",
      "role": "Chief Technology Officer",
      "company": "Ironclad",
      "twitter": "",
      "talks": [
        {
          "title": "Dealing with AI's Cost Problem without Sacrificing Innovation",
          "description": "AI adoption is accelerating, but the economics are starting to strain. This week, one company (rumored to be Amazon AWS) spent half a billion on AI in a single month after failing to put usage limits on Claude for employees. This is an extreme cause, but the sentiment remains. Organizations are swinging from tokenmaxxing to AI efficiency. Sunita can connect what is happening at the infrastructure level to what enterprises are doing in practice including early moves toward on-prem deployments and more selective use of AI in production. She'll share examples she is seeing of where companies are successfully scaling back AI spend while still using AI to add value.",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05785037,
        0.051151752,
        0.010769902,
        0.014896075,
        0.06303166,
        -0.022186646,
        -0.0061447057,
        -0.016628943,
        0.08455576,
        -0.18227689,
        0.00018310332,
        -0.09079583,
        0.0022048776,
        -0.05740932,
        -0.04497513,
        0.0108068725,
        0.02812741,
        -0.032244403,
        -0.10233918,
        -0.05697373,
        -0.008840345,
        -0.022275954,
        -0.024266502,
        0.034891024,
        -0.034247857,
        0.1005903,
        0.023876691,
        -0.04387254,
        -0.094061315,
        0.45823067,
        -0.09487846,
        -0.016560763,
        0.04193069,
        0.00712521,
        0.023909193,
        -0.095049955,
        0.03282659,
        -0.009766266,
        0.011925685,
        0.06068686,
        -0.02063664,
        0.07751663,
        -0.052158825,
        0.032689933,
        -0.013970558,
        -0.061501317,
        -0.035030082,
        -0.0065460973,
        -0.061690193,
        0.03465485,
        -0.10075327,
        -0.02707324,
        0.01491911,
        0.021110257,
        0.044071797,
        0.0066069462,
        0.026245087,
        -0.09556721,
        0.094324216,
        -0.03333994,
        -0.17978817,
        -0.0119279185,
        -0.027846908,
        0.047280744,
        -0.017628556,
        -0.082416266,
        0.04894166,
        -0.004219815,
        -0.031258248,
        -0.014151848,
        -0.019476913,
        -0.030081676,
        -0.07164269,
        -0.015564782,
        -0.031264815,
        -0.13321635,
        -0.049214453,
        0.10165523,
        0.024557153,
        0.041250702,
        -0.017026788,
        0.06648248,
        -0.14663187,
        -0.0061196773,
        0.09682599,
        -0.036394592,
        -0.094303705,
        0.04017986,
        -0.045406297,
        0.04076564,
        -0.0048711086,
        -0.052505374,
        -0.035045065,
        0.01291549,
        -0.104305126,
        0.056644928,
        -0.06714871,
        0.016132066,
        -0.03727091,
        0.019526964,
        0.012781358,
        0.0023527306,
        -0.0007648324,
        0.016765928,
        -0.015087112,
        -0.019902091,
        0.024839925,
        -0.0047718226,
        0.026160246,
        0.02115818,
        -0.069079526,
        -0.60554355,
        0.093994595,
        0.06127482,
        -0.05465501,
        -0.16270287,
        -0.0130084595,
        -0.055757716,
        0.078877956,
        0.07223516,
        -0.04340018,
        0.03084516,
        -0.026996024,
        -0.0057349964,
        -0.011888661,
        0.05488458,
        0.004148736,
        0.065970875
      ]
    },
    {
      "id": "worldsfair-speaker-216",
      "conference": "worldsfair",
      "name": "Simran Arora",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Can LLMs write fast multi-GPU kernels? We built a benchmark to find out.",
          "description": "LLMs have gotten surprisingly good at writing GPU kernels, but almost all the benchmarks measuring that progress are single-GPU. In production, communication is the bottleneck: all-reduce alone accounts for over 20% of inference latency on Llama-3.3-70B, and that gap keeps widening as compute scales faster than interconnect bandwidth. ParallelKernelBench (PKB) offers a benchmark and evaluation framework for multi-GPU kernel generation and includes 87 problems from real codebases where the task is replacing PyTorch + NCCL with a CUDA kernel that moves data directly over NVLink. We tested GPT-5.5, Gemini 3 Pro, Opus 4.7, and other frontier coding models. Under a third of problems solved were correctly, and fewer than a quarter of those beat the naive baseline. We'll cover why they fail, what the patterns look like, and a few cases where models produced kernels faster than anything publicly available, including one for NVIDIA NeMo-RL's GRPO training loop, which has no prior optimized public reference. The benchmark is open source and we want to see what you can do!",
          "day": "Day 3 — Session Day 2",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.069545925,
        0.09990909,
        0.013935628,
        0.08133574,
        0.04064538,
        -0.032586373,
        -0.065604396,
        0.0021834674,
        0.013417046,
        -0.21239886,
        0.0059660715,
        0.0029011178,
        0.020365087,
        -0.014248559,
        -0.03523911,
        -0.008970904,
        0.058791526,
        -0.0658815,
        -0.074941546,
        -0.091249004,
        0.009757264,
        -0.13920778,
        0.014113666,
        0.0618279,
        -0.06667798,
        0.060031846,
        0.0046537165,
        -0.024806337,
        -0.09501086,
        0.4065279,
        -0.091877006,
        0.017694075,
        -0.010006781,
        -0.085602395,
        0.08745822,
        -0.12647375,
        0.05555824,
        0.030574473,
        0.064430386,
        0.044527903,
        0.033981103,
        0.03324963,
        -0.0643332,
        0.076915994,
        -0.07145926,
        -0.05298144,
        -0.05536741,
        -0.018569548,
        0.014006696,
        -0.05898454,
        -0.06714975,
        -0.045338467,
        0.021943768,
        -0.028246019,
        0.020940596,
        0.07790135,
        -0.03553953,
        -0.072917566,
        0.12528016,
        0.035445828,
        -0.104794316,
        0.016294742,
        0.067674264,
        0.03272702,
        0.002537367,
        -0.011180415,
        0.059370294,
        -0.08845468,
        -0.02266861,
        -0.014110051,
        0.09357433,
        -0.021118201,
        -0.0407411,
        0.028293207,
        -0.019766256,
        -0.1621184,
        -0.032376304,
        0.08394707,
        -0.012963332,
        0.07629785,
        -0.014126795,
        0.04010653,
        -0.00066323834,
        -0.028949305,
        0.046581756,
        0.054903768,
        -0.04832603,
        0.081842974,
        -0.055662062,
        0.051084336,
        0.02950989,
        0.00012811628,
        -0.07315658,
        0.003094138,
        0.010475719,
        0.029276177,
        -0.01692048,
        0.058005195,
        -0.042565867,
        0.031133609,
        0.07406008,
        0.003217958,
        0.05618627,
        -0.028686091,
        -0.03930392,
        -0.03978297,
        0.065507084,
        -0.008457816,
        -0.035699334,
        -0.06120218,
        -0.036014896,
        -0.62590903,
        -0.0033806753,
        0.009656579,
        -0.025858603,
        -0.1680115,
        0.044572838,
        -0.016579386,
        -0.0641995,
        0.027632583,
        -0.036599748,
        0.06408067,
        0.01523115,
        -0.01188461,
        -0.039121613,
        -0.043772712,
        0.03565069,
        0.03511855
      ]
    },
    {
      "id": "worldsfair-speaker-217",
      "conference": "worldsfair",
      "name": "Tejas Bhakta",
      "role": "",
      "company": "Morph",
      "twitter": "https://x.com/tejasybhakta",
      "talks": [
        {
          "title": "Autoresearch for Kernels",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.032697648,
        0.04842795,
        0.049606547,
        0.051715244,
        0.07316808,
        -0.0145280985,
        -0.076868884,
        0.009053871,
        0.024165442,
        -0.21743177,
        0.0065419516,
        0.021205135,
        -0.039799727,
        0.014104707,
        -0.024813337,
        0.012598271,
        0.07117731,
        -0.08389981,
        -0.08316278,
        -0.051848445,
        0.015927855,
        -0.09336397,
        -0.066067174,
        0.03664384,
        -0.017624278,
        0.009467289,
        -0.004569921,
        0.0028558346,
        -0.15920669,
        0.44314328,
        -0.07340742,
        -0.05474708,
        0.008622022,
        -0.08468206,
        0.06338025,
        -0.011985673,
        0.023556747,
        0.054120593,
        0.0068842703,
        -0.050718326,
        -0.07338115,
        0.04226348,
        0.016324557,
        0.0351961,
        -0.022326956,
        0.026106406,
        -0.06334438,
        -0.061971623,
        -0.030616246,
        0.011971908,
        -0.11461593,
        0.042391922,
        0.022774411,
        -0.065403,
        0.025156196,
        -0.013452262,
        -0.00036564318,
        -0.05238869,
        0.096517116,
        -0.03908567,
        -0.058932677,
        -0.08122417,
        0.012414375,
        -0.0015397185,
        -0.07711036,
        0.0031150861,
        0.09142005,
        -0.13675544,
        0.016699815,
        0.036998875,
        0.04352078,
        0.027864067,
        0.0055790567,
        0.07962171,
        -0.09425167,
        -0.02444115,
        -0.02793161,
        0.13052604,
        -0.010896182,
        0.012597919,
        0.028737405,
        0.019728502,
        -0.024685986,
        -0.074007064,
        -0.025415476,
        -0.034275625,
        -0.05458895,
        -0.004412167,
        0.063143365,
        0.04058786,
        0.07669622,
        0.046143405,
        -0.019320684,
        -0.017190523,
        -0.04887621,
        0.049459126,
        -0.07707581,
        0.071584925,
        0.037598215,
        -0.017798299,
        -0.059391472,
        0.021674003,
        0.0585192,
        -0.02847647,
        -0.032015983,
        -0.03728315,
        -0.003992244,
        -0.018423311,
        -0.0062947175,
        -0.008338301,
        -0.052621566,
        -0.6380043,
        0.012355086,
        0.010880629,
        0.009509965,
        -0.149701,
        -0.015088924,
        0.011670528,
        0.10175133,
        0.0568982,
        0.0054076565,
        0.030938147,
        -0.01883136,
        0.065977424,
        0.020825662,
        -0.014352657,
        0.028222794,
        0.03557554
      ]
    },
    {
      "id": "worldsfair-speaker-218",
      "conference": "worldsfair",
      "name": "Adi Singh",
      "role": "Co-founder",
      "company": "AgentMail",
      "twitter": "",
      "talks": [
        {
          "title": "The Next Trillion Users of the Internet Still Don't Have an Identity",
          "description": "In the last few months, hundreds of thousands of people set up personal AI agents that send email on their behalf, manage calendars, book travel, even sign contracts - all thanks to openclaw. Most of these agents have no real identity online. They borrow a human's. The identity stack of the internet, OAuth, 2FA, KYC, magic links, was built for people sitting at a keyboard. Agents don't fit, and we've ended up with shared accounts, hard-coded credentials, and humans dragged back into every loop. I'm Adi, co-founder of AgentMail. We are building the identity layer for what we believe will be the next trillion users of the internet, and they will not be human. Across hundreds of customers, we have watched what breaks when an agent has no real address. It fails at signups. Verification codes get lost. There is no accountability when something goes wrong. The human gets pulled back in. This talk is the case for making agents first-class citizens of the internet. I'll cover the identity architecture we've shipped, the legacy industries already adopting it and making real money, and where agent identity infrastructure is going over the next decade.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.027044313,
        0.09225893,
        0.04512622,
        -0.021788428,
        0.034148846,
        -0.049863562,
        -0.06355675,
        -0.020714996,
        0.02002976,
        -0.24935406,
        0.027148027,
        -0.029447619,
        -0.021701595,
        -0.098146945,
        -0.07380062,
        0.037009057,
        0.08466444,
        -0.03971465,
        -0.04435328,
        -0.10625232,
        -0.041260973,
        -0.079903275,
        -0.0068773692,
        0.014008498,
        -0.048161358,
        0.10420569,
        -0.0006318726,
        0.020305807,
        -0.08852456,
        0.45524883,
        -0.027597439,
        0.02294337,
        0.024143325,
        -0.020552466,
        0.06200382,
        -0.073290825,
        -0.015774406,
        -0.019191522,
        0.02646117,
        0.091131896,
        -0.046078257,
        0.03787748,
        -0.041907378,
        0.046434056,
        0.013520043,
        -0.044024184,
        -0.03742113,
        -0.01735921,
        0.035221882,
        -0.016488891,
        -0.011323213,
        0.017535463,
        0.060970634,
        0.025529739,
        0.051267497,
        0.05100506,
        0.030475013,
        -0.08195446,
        0.07986436,
        -0.03863935,
        -0.061073046,
        -0.00024042769,
        -0.008763945,
        0.046898946,
        0.0014149969,
        -0.12585601,
        0.0329203,
        -0.06530272,
        -0.0022360266,
        -0.05166865,
        -0.031012036,
        0.055635292,
        -0.04022147,
        0.044944227,
        -0.064080834,
        -0.11507807,
        -0.056399778,
        0.03361812,
        -0.015595467,
        0.039221868,
        -0.020565268,
        0.058688868,
        -0.07876747,
        -0.05205411,
        -0.015514207,
        -0.08788761,
        -0.0468934,
        0.043732524,
        -0.0075937067,
        0.05760559,
        0.007480746,
        -0.029058177,
        -0.01907598,
        -0.015717097,
        -0.101003125,
        0.015373391,
        -0.04101168,
        0.02931273,
        0.04288939,
        0.02193692,
        0.046231955,
        -0.05169615,
        0.03347959,
        0.042588405,
        0.014829811,
        -0.011479461,
        0.093666665,
        -0.0038011458,
        0.038805533,
        -0.04822735,
        -0.021851983,
        -0.6355836,
        0.021934476,
        0.044812113,
        -0.05281068,
        -0.15038992,
        0.023365114,
        0.008759763,
        0.0015659076,
        0.07308411,
        -0.058979325,
        0.06806266,
        0.022005213,
        0.026810866,
        -0.05592405,
        0.029378908,
        0.0415203,
        0.020467248
      ]
    },
    {
      "id": "worldsfair-speaker-219",
      "conference": "worldsfair",
      "name": "Cormac Brick",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Why Large? Tiny LMs & Agents on Edge/Robotics",
          "description": "big models get a lot of press. small model scale much better. RAM is expensive. The real world needs tiny models for scale on the edge. This workshop will cover how to combine both for mobile and robotics deployment. specifically covering: - skills are different on mobile - tiny LLMs <1B scale much further on mobile/web - how to fine tune and train tiny models. - skills on robotics / edge/ mobile - latest open models for edge (including gemma, qwen, and anything else that happens in next 10 weeks) This talk will focus on open models, including some gemma variants that will be shortly announced.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.00792043,
        0.028997784,
        -0.008827783,
        0.057637203,
        0.044976342,
        -0.0048685875,
        -0.09101508,
        -0.0060814493,
        0.04645248,
        -0.2069329,
        0.0073109786,
        -0.05174672,
        0.031869046,
        -0.0379035,
        -0.07333489,
        0.008627728,
        0.098407686,
        -0.0748612,
        -0.078447044,
        -0.03490542,
        -0.0087676,
        -0.075793,
        0.044651248,
        0.048281483,
        -0.05121383,
        0.04503734,
        -0.05483611,
        0.06787587,
        -0.07997646,
        0.47552654,
        -0.10676595,
        -0.038025737,
        0.067740016,
        -0.019858941,
        0.022237057,
        -0.0574367,
        -0.07069352,
        -0.02297965,
        0.065854244,
        0.012713807,
        -0.020274432,
        0.0028009426,
        -0.006684267,
        0.041758597,
        -0.05583179,
        0.05657968,
        -0.058019776,
        0.0018634916,
        -0.015911285,
        -0.015763728,
        -0.09959476,
        0.038033325,
        0.033041917,
        0.05567606,
        0.062055726,
        0.033996504,
        -0.017260652,
        -0.010387024,
        0.11096539,
        -0.04024296,
        -0.055943925,
        -0.05484807,
        -0.059104204,
        -0.047537543,
        -0.050378636,
        0.0023441976,
        0.07039057,
        -0.07415709,
        0.027681988,
        -0.046263233,
        0.022402676,
        -0.022109183,
        -0.020164292,
        -0.017599551,
        -0.045282885,
        -0.089126006,
        -0.087773725,
        0.06384352,
        0.03064765,
        -0.04948038,
        -0.05122964,
        0.038392454,
        -0.039125305,
        -0.03605532,
        0.021949276,
        -0.09624758,
        -0.07654398,
        0.075772926,
        0.028802019,
        0.054343812,
        0.07347866,
        0.006951903,
        -0.08844517,
        -0.019136647,
        -0.08080971,
        0.0010802097,
        0.003660082,
        0.053593762,
        -0.023390273,
        -0.024117328,
        -0.054231282,
        -0.028771993,
        0.046001233,
        0.016865905,
        -0.067729235,
        -0.066481225,
        0.04474262,
        0.010569363,
        0.0030558493,
        -0.03854613,
        -0.009115872,
        -0.64561385,
        -0.028876493,
        0.031877298,
        -0.063931674,
        -0.122068584,
        0.009077557,
        0.0023865746,
        -0.013404193,
        0.029750591,
        -0.018911628,
        0.027243283,
        0.026895992,
        -0.039319247,
        -0.019348275,
        0.017439593,
        0.023297537,
        0.0019374825
      ]
    },
    {
      "id": "worldsfair-speaker-220",
      "conference": "worldsfair",
      "name": "Sara Hooker",
      "role": "Co-founder",
      "company": "Adaption Labs",
      "twitter": "https://x.com/sarahookr",
      "talks": [
        {
          "title": "Adaption Labs — Gradient-Free Continual Learning",
          "description": "Gradient-free continual learning for AI systems that adapt from real-world experience.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.06309915,
        0.019897552,
        0.09044255,
        -0.0025397423,
        0.035711087,
        -0.043657873,
        -0.07837082,
        -0.07837013,
        0.101622246,
        -0.23254341,
        -0.00054667087,
        -0.038690183,
        -0.014632805,
        -0.032308966,
        -0.05096532,
        0.02861749,
        0.1392785,
        -0.091835596,
        -0.028357195,
        -0.04559518,
        -0.048002083,
        -0.09508725,
        -0.016185766,
        0.09247556,
        -0.097445674,
        0.11657206,
        0.0057953475,
        -0.05567353,
        -0.08328588,
        0.42102593,
        -0.10417406,
        -0.029866414,
        0.025764383,
        0.014318242,
        0.015284221,
        -0.061628077,
        -0.01852705,
        0.034113105,
        -0.007866266,
        0.050977897,
        -0.029387748,
        -0.03285425,
        -0.016505305,
        0.04642381,
        -0.07473288,
        -0.07149267,
        -0.059396375,
        -0.07438713,
        -0.020849125,
        0.067750044,
        -0.10857309,
        0.0028119863,
        0.07006314,
        0.011923057,
        0.045556158,
        0.060650807,
        -0.027090168,
        -0.07028968,
        0.065783836,
        -0.008286057,
        -0.096411094,
        -0.0341377,
        -0.027794447,
        0.06062462,
        -0.015010353,
        -0.04293206,
        0.07842817,
        -0.045072354,
        -0.008286677,
        -0.011563689,
        0.030691322,
        -0.016181612,
        -0.006321411,
        0.020213785,
        -0.12604137,
        -0.12768695,
        -0.002899521,
        0.0023242,
        0.014351871,
        -0.021537514,
        -0.015103313,
        0.08794991,
        -0.08481018,
        -0.03736875,
        0.05078469,
        -0.03778616,
        -0.062141024,
        0.005122979,
        -0.03586444,
        0.02739709,
        0.036145598,
        0.0070961653,
        -0.0066579236,
        -0.027622776,
        -0.03569106,
        0.01145812,
        -0.09491864,
        0.08879648,
        0.060842276,
        0.059342097,
        -0.019845225,
        -0.08135096,
        0.05418856,
        -0.059033226,
        -0.032231435,
        -0.058870193,
        -0.0436341,
        -0.035216685,
        0.058529813,
        0.013391891,
        -0.025148403,
        -0.60164434,
        -0.03810808,
        -0.0031594224,
        -0.0051485053,
        -0.18918382,
        -0.027102457,
        -0.026161436,
        0.0040946603,
        0.090560436,
        -0.034811385,
        0.022709684,
        -0.012758502,
        0.06467658,
        0.027605942,
        -0.022072157,
        0.04211501,
        -0.0100088585
      ]
    },
    {
      "id": "worldsfair-speaker-221",
      "conference": "worldsfair",
      "name": "Brian Douglas",
      "role": "Co-founder and CEO",
      "company": "Paper Compute Company",
      "twitter": "https://x.com/bdougieYO",
      "talks": [
        {
          "title": "Don't Write Skills, Train Models",
          "description": "Every AI agent call generates training data. Most teams throw it away. They write skills files instead. Text documents that describe how to do a task and hope the model follows them at inference time. Skills work until they don't. The model drifts, skips steps, hallucinates a shortcut. So you rewrite the skill, add more constraints, hope harder. There's a better path. If you've used a skill enough to know what good output looks like, you already have training data. You just aren't using it. This talk covers what I learned building an open source fine-tuning pipeline that turns agent session traces into SFT and DPO training datasets. A telemetry proxy captures every LLM call as a content-addressed Merkle DAG with zero instrumentation. Successful sessions become supervised fine-tuning data. Pair them against failures, matched by goal category, and you get preference pairs for DPO. No manual labeling. No synthetic data. But training data quality depends on environment consistency. If the same agent produces different results because of package drift, nondeterministic toolchains, or inconsistent system state, your training signal is noise. This is where NixOS changes the equation. A hardened, reproducible OS means every agent session runs against an identical, declarative environment. Nix controls the variables that sandboxing alone doesn't: dependency graphs, system libraries, toolchain versions. When you can guarantee the environment is the same across hundreds of sessions, the behavioral signal in your traces is actually trustworthy. We'll walk through the full pipeline. How to rebuild parent-hash chains from a SQLite database and join facet metadata. How to filter to fully_achieved sessions and truncate 82k-token conversations down to 4k-6k training examples using summary context plus the last three turns. How to match success/failure pairs by goal category and exclude unclear_requirements failures so DPO learns from real agent mistakes, not ambiguous prompts. How QLoRA keeps VRAM low enough to train a 7B model on a single consumer GPU. And what happens when you try DPO on 12GB VRAM (two simultaneous forward passes for logprob computation will teach you about gradient accumulation settings fast). The result: a LoRA adapter trained on your own agent traces, in a reproducible environment, on a single consumer GPU, for less than $2 in cloud compute. No YAML. One config file. All code is open source.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        },
        {
          "title": "Don't Write Skills, Train Models (cont. 2/3)",
          "description": "Continuation block 2 of 3 for Brian Douglas's workshop session.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        },
        {
          "title": "Don't Write Skills, Train Models (cont. 3/3)",
          "description": "Continuation block 3 of 3 for Brian Douglas's workshop session.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.034930993,
        0.0033642524,
        0.022274707,
        -0.053663436,
        -0.010008722,
        -0.03130134,
        -0.07605055,
        -0.014070614,
        0.016230451,
        -0.23125532,
        -0.029270647,
        -0.037679575,
        0.0027101182,
        -0.05491782,
        -0.03474485,
        -0.0011337157,
        0.02556476,
        -0.05868793,
        -0.03748011,
        -0.013924899,
        0.005268123,
        -0.03661928,
        -0.04584522,
        0.051864494,
        -0.0089457175,
        0.109743126,
        0.0122496,
        0.06820713,
        -0.02031677,
        0.40975833,
        -0.11351785,
        -0.014747286,
        -0.0033978398,
        0.00006157056,
        0.0720764,
        -0.032874804,
        -0.038970973,
        0.031256787,
        -0.019303475,
        0.021743042,
        -0.05339306,
        0.030338973,
        -0.018313605,
        0.048980497,
        -0.008619269,
        -0.04213909,
        -0.048410974,
        -0.014745667,
        0.059383456,
        -0.032539282,
        -0.025832022,
        0.013634915,
        0.065464675,
        0.050045837,
        0.028299544,
        0.028635403,
        -0.0010457768,
        -0.13658938,
        0.12043544,
        -0.009685098,
        -0.084877856,
        0.026271457,
        0.03394317,
        0.026424382,
        0.004628768,
        -0.053313743,
        0.040303495,
        -0.029784942,
        -0.031570483,
        -0.13369256,
        -0.03839986,
        0.045186866,
        -0.038810622,
        0.042918965,
        -0.1263623,
        -0.115185864,
        -0.04164378,
        0.076842785,
        -0.037500557,
        -0.008130805,
        -0.0036173237,
        0.045140654,
        -0.06882391,
        -0.11603566,
        0.013531677,
        -0.09122744,
        -0.08563192,
        0.012727455,
        0.0011031573,
        0.08357001,
        0.039003648,
        -0.015699225,
        -0.070824,
        -0.022976317,
        -0.072407834,
        0.009606123,
        -0.0363055,
        0.045660876,
        -0.0321698,
        -0.010704417,
        -0.011979264,
        0.008935806,
        0.059223007,
        -0.004544801,
        0.05017732,
        -0.09783293,
        -0.059906792,
        -0.041720696,
        0.06035375,
        -0.030844348,
        -0.036929827,
        -0.657019,
        -0.009285224,
        -0.0382373,
        -0.045449696,
        -0.14148808,
        0.032808237,
        -0.0002968785,
        0.053642515,
        0.05920534,
        0.07721552,
        0.057164565,
        0.052164484,
        0.0001800084,
        -0.008050789,
        0.0790629,
        -0.00009966804,
        0.04928465
      ]
    },
    {
      "id": "worldsfair-speaker-222",
      "conference": "worldsfair",
      "name": "John McBride",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Don't Write Skills, Train Models",
          "description": "Every AI agent call generates training data. Most teams throw it away. They write skills files instead. Text documents that describe how to do a task and hope the model follows them at inference time. Skills work until they don't. The model drifts, skips steps, hallucinates a shortcut. So you rewrite the skill, add more constraints, hope harder. There's a better path. If you've used a skill enough to know what good output looks like, you already have training data. You just aren't using it. This talk covers what I learned building an open source fine-tuning pipeline that turns agent session traces into SFT and DPO training datasets. A telemetry proxy captures every LLM call as a content-addressed Merkle DAG with zero instrumentation. Successful sessions become supervised fine-tuning data. Pair them against failures, matched by goal category, and you get preference pairs for DPO. No manual labeling. No synthetic data. But training data quality depends on environment consistency. If the same agent produces different results because of package drift, nondeterministic toolchains, or inconsistent system state, your training signal is noise. This is where NixOS changes the equation. A hardened, reproducible OS means every agent session runs against an identical, declarative environment. Nix controls the variables that sandboxing alone doesn't: dependency graphs, system libraries, toolchain versions. When you can guarantee the environment is the same across hundreds of sessions, the behavioral signal in your traces is actually trustworthy. We'll walk through the full pipeline. How to rebuild parent-hash chains from a SQLite database and join facet metadata. How to filter to fully_achieved sessions and truncate 82k-token conversations down to 4k-6k training examples using summary context plus the last three turns. How to match success/failure pairs by goal category and exclude unclear_requirements failures so DPO learns from real agent mistakes, not ambiguous prompts. How QLoRA keeps VRAM low enough to train a 7B model on a single consumer GPU. And what happens when you try DPO on 12GB VRAM (two simultaneous forward passes for logprob computation will teach you about gradient accumulation settings fast). The result: a LoRA adapter trained on your own agent traces, in a reproducible environment, on a single consumer GPU, for less than $2 in cloud compute. No YAML. One config file. All code is open source.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05008088,
        0.07102086,
        0.06736464,
        -0.0062350594,
        0.013763375,
        -0.030933427,
        -0.06439342,
        0.0032141558,
        0.024187673,
        -0.2620603,
        -0.008852881,
        -0.01394303,
        0.016808344,
        -0.043583456,
        -0.020583145,
        0.023234459,
        0.108706996,
        -0.086460516,
        -0.09853779,
        -0.07605116,
        -0.043382425,
        -0.10749919,
        -0.01512119,
        0.0631361,
        -0.02128825,
        0.07396846,
        0.0042970283,
        0.06372542,
        -0.05784004,
        0.43437135,
        -0.08885434,
        -0.0131907705,
        0.0011962862,
        0.004171508,
        0.0472315,
        -0.029590823,
        -0.0019846119,
        0.032187402,
        -0.040166054,
        0.07670893,
        -0.05398409,
        0.0507053,
        -0.03355993,
        0.028782953,
        -0.01091711,
        0.010646381,
        -0.08415845,
        -0.0051450515,
        0.019221094,
        -0.014968659,
        -0.080894955,
        -0.017119477,
        0.063333444,
        0.040163666,
        0.00931942,
        0.058290232,
        0.016424635,
        -0.055219494,
        0.11848148,
        -0.074168816,
        -0.069319144,
        0.0017894015,
        -0.014495592,
        -0.002267543,
        0.04625343,
        -0.036924563,
        0.084237576,
        -0.06208059,
        -0.025646,
        -0.09231196,
        -0.06793172,
        0.046954736,
        -0.041761752,
        0.06346212,
        -0.11823506,
        -0.09035852,
        -0.025350645,
        0.07941864,
        -0.027240843,
        0.016986182,
        -0.026082395,
        0.09256882,
        -0.036652967,
        -0.055260245,
        -0.05810059,
        -0.04930971,
        -0.086760685,
        0.041841295,
        -0.0016183418,
        0.06724053,
        0.07550829,
        -0.028968003,
        -0.05907342,
        0.002436048,
        -0.028452436,
        -0.06435065,
        -0.004557828,
        0.039166793,
        -0.015048404,
        0.044184476,
        -0.0280799,
        0.009597882,
        0.041193105,
        0.0065413113,
        0.0024221232,
        -0.03600533,
        -0.009311669,
        -0.03654676,
        0.04418984,
        -0.008703034,
        -0.02570446,
        -0.63768756,
        0.002533704,
        -0.05980483,
        -0.016769227,
        -0.13617407,
        -0.0040156795,
        0.023348197,
        -0.0051501454,
        0.05228663,
        0.028728455,
        0.09485954,
        0.009975945,
        -0.0124444645,
        -0.0034773587,
        0.0597887,
        0.05767071,
        0.027652327
      ]
    },
    {
      "id": "worldsfair-speaker-223",
      "conference": "worldsfair",
      "name": "Cornelia Davis",
      "role": "Sr. Staff Developer Advocate",
      "company": "Temporal",
      "twitter": "https://x.com/cdavisafc",
      "talks": [
        {
          "title": "MCP Tasks (async)/ Why the heck aren't any agents supporting MCP tasks/async?",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0027515956,
        0.04964574,
        0.023831492,
        -0.02400579,
        0.032044623,
        -0.010282523,
        -0.05785458,
        -0.111579515,
        0.08066214,
        -0.17564595,
        -0.019162048,
        -0.026518416,
        0.000083458806,
        -0.06307555,
        -0.0011536044,
        -0.012015195,
        0.061547555,
        -0.061975148,
        -0.035086952,
        -0.13548054,
        -0.003082033,
        -0.05326354,
        0.003119036,
        0.04747894,
        -0.04665619,
        0.042047102,
        -0.052585613,
        0.022909971,
        -0.11473913,
        0.45614377,
        -0.098060735,
        0.007503964,
        0.070232674,
        -0.030475274,
        0.09520357,
        -0.034861427,
        -0.0307966,
        0.003307007,
        0.030966537,
        -0.002354558,
        -0.06786789,
        0.018482193,
        -0.032978706,
        0.0995542,
        -0.1783121,
        -0.035950415,
        -0.09199245,
        0.011521638,
        -0.013596819,
        0.0077234562,
        -0.022200517,
        -0.0036777037,
        0.050752625,
        -0.014668637,
        0.04638381,
        -0.013426992,
        -0.0030422069,
        -0.07073692,
        -0.000033496875,
        -0.012342831,
        -0.0909852,
        0.011887048,
        -0.032786362,
        -0.000466395,
        0.02225626,
        -0.044929154,
        0.07244709,
        -0.025396239,
        0.030544477,
        -0.061621375,
        0.07689814,
        0.029856997,
        -0.031930473,
        0.023241278,
        -0.09551821,
        -0.05174934,
        -0.01700437,
        0.061613612,
        -0.07759884,
        0.03510853,
        -0.010518311,
        0.10541503,
        -0.08921597,
        -0.13436654,
        -0.022441473,
        -0.015916578,
        -0.060100395,
        -0.01327747,
        -0.094106406,
        -0.014767191,
        0.024726065,
        -0.012297623,
        0.006811282,
        0.01870538,
        -0.0613918,
        0.014826598,
        -0.058930952,
        -0.015834207,
        -0.034662936,
        0.037521288,
        -0.05777853,
        -0.07631952,
        0.010231657,
        -0.02343633,
        0.046419885,
        -0.030703964,
        -0.050585948,
        0.01053568,
        0.1277833,
        -0.05773284,
        -0.045648832,
        -0.60123014,
        -0.030052103,
        -0.03977654,
        -0.047349054,
        -0.16690032,
        0.021588825,
        -0.021625698,
        0.07699373,
        0.086498536,
        0.021118963,
        0.090284415,
        -0.03615803,
        -0.0032886018,
        0.042539928,
        0.009502563,
        0.07279946,
        0.019474436
      ]
    },
    {
      "id": "worldsfair-speaker-224",
      "conference": "worldsfair",
      "name": "Uri Rolls",
      "role": "CEO",
      "company": "Arithmetic",
      "twitter": "",
      "talks": [
        {
          "title": "Arithmetic (fka Operating Intelligence)",
          "description": "Hold for a startup referred to as Operating Intelligence. Web search did not confidently resolve a single company, so this hold is intentionally marked TBD and should be refined before publish.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.052274294,
        0.0112719545,
        0.029549371,
        0.027861977,
        -0.011682912,
        -0.003942447,
        -0.08738576,
        0.027293578,
        0.035487067,
        -0.25006518,
        -0.001545678,
        -0.034228787,
        -0.0071943537,
        -0.012851629,
        -0.05411851,
        -0.025538037,
        0.041113686,
        0.0058969185,
        -0.019168943,
        -0.08269407,
        -0.0068603866,
        -0.0160484,
        -0.03884934,
        0.04629038,
        -0.026704926,
        0.05000441,
        0.066609636,
        0.047977954,
        -0.124662325,
        0.44853315,
        -0.07602304,
        -0.0460492,
        -0.055881456,
        0.06357357,
        0.08383695,
        -0.032000493,
        -0.037401903,
        -0.06263315,
        -0.016813021,
        0.052465647,
        -0.05426796,
        -0.0026235282,
        -0.06316494,
        0.016238188,
        0.018782951,
        -0.058139008,
        -0.08922043,
        0.016839322,
        -0.0012201705,
        0.02832957,
        -0.0011370245,
        0.030921921,
        0.039651304,
        0.09681588,
        0.07236896,
        0.027194275,
        -0.044510577,
        -0.06792307,
        0.07967441,
        -0.053960986,
        -0.06950212,
        0.07059236,
        0.010919383,
        0.010081997,
        0.052667588,
        -0.063167274,
        0.10852273,
        -0.056220666,
        0.021915711,
        0.032087915,
        0.016564384,
        0.070806645,
        -0.11635595,
        -0.046428733,
        -0.062593564,
        -0.08147867,
        -0.039243672,
        0.07609428,
        0.02860297,
        0.014444092,
        0.037702497,
        0.07304648,
        -0.025582625,
        0.010346031,
        0.054062355,
        -0.053299963,
        -0.06500178,
        0.07440864,
        0.034796055,
        0.09315764,
        0.032469917,
        -0.02782948,
        0.007294392,
        0.019554045,
        -0.074764706,
        -0.014579362,
        -0.040207036,
        0.07491526,
        -0.035428043,
        -0.031116396,
        -0.00045928863,
        0.032604635,
        0.06538263,
        -0.023251247,
        -0.03466073,
        -0.08534804,
        0.021032952,
        0.046347752,
        0.013797957,
        0.012469614,
        -0.042659607,
        -0.6005433,
        0.10151878,
        -0.019679999,
        -0.11302201,
        -0.15230046,
        0.030250382,
        0.014428776,
        0.058797576,
        0.100304656,
        -0.011281071,
        0.118633784,
        -0.020802317,
        0.041674744,
        0.017935913,
        0.061870907,
        0.063017264,
        -0.06536475
      ]
    },
    {
      "id": "worldsfair-speaker-225",
      "conference": "worldsfair",
      "name": "Roberto Milev",
      "role": "",
      "company": "Navan",
      "twitter": "",
      "talks": [
        {
          "title": "Agents Are Where Microservices Were in 2015. We're Making All the Same Mistakes.",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.033074085,
        0.03223331,
        -0.0051003355,
        -0.010440408,
        -0.054173335,
        0.010180637,
        -0.07897979,
        0.014573794,
        0.02340616,
        -0.25994653,
        -0.082604796,
        -0.005755787,
        0.09856449,
        0.02979255,
        -0.072304174,
        -0.01482749,
        0.087024376,
        -0.05400981,
        -0.067466624,
        -0.20179784,
        -0.09165482,
        -0.09715403,
        0.016759578,
        0.033796396,
        0.038436256,
        0.091461204,
        -0.017083816,
        0.05503057,
        -0.11473715,
        0.4879836,
        -0.0848601,
        0.0027621023,
        0.0646928,
        -0.04439236,
        -0.03314962,
        -0.058059722,
        -0.08552573,
        0.045360904,
        -0.04616476,
        0.11148684,
        -0.035495568,
        0.026323257,
        -0.04941562,
        -0.006807777,
        -0.056835406,
        0.07386868,
        -0.021597318,
        -0.016211377,
        0.015701046,
        -0.04776173,
        -0.10925121,
        0.06587553,
        0.03854224,
        0.004753599,
        0.040349994,
        -0.023884859,
        -0.01292737,
        -0.067127794,
        0.030059738,
        -0.006707084,
        -0.07402441,
        0.010339241,
        -0.05941737,
        -0.028321432,
        -0.037781533,
        -0.06394522,
        0.10188349,
        -0.07894781,
        0.04928933,
        -0.020574551,
        0.0034426155,
        0.034107156,
        -0.084408805,
        0.018038098,
        -0.05026588,
        -0.056041215,
        -0.03904167,
        0.10816305,
        -0.034965955,
        0.021980226,
        0.024274895,
        0.09358193,
        -0.06671107,
        -0.015194965,
        0.07416017,
        0.012883823,
        -0.013422751,
        0.014222471,
        0.027288338,
        0.06833739,
        0.064307265,
        -0.022421299,
        -0.06486168,
        0.0052420427,
        -0.077865876,
        0.026037438,
        -0.028943414,
        0.109996185,
        -0.065612085,
        0.057992455,
        -0.018227672,
        -0.0353848,
        0.0004864155,
        -0.016344976,
        -0.00879638,
        -0.012252917,
        -0.008975619,
        -0.00411264,
        0.07450839,
        -0.10001988,
        0.014964445,
        -0.527676,
        -0.023794563,
        -0.014831957,
        -0.011599004,
        -0.1604034,
        0.05732844,
        0.067706816,
        0.02661875,
        0.07010136,
        -0.050355397,
        -0.0041094576,
        -0.048120923,
        -0.028088596,
        0.058013752,
        -0.022464775,
        -0.0030384574,
        -0.04546533
      ]
    },
    {
      "id": "worldsfair-speaker-226",
      "conference": "worldsfair",
      "name": "Uday Kanagala",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Agents Are Where Microservices Were in 2015. We're Making All the Same Mistakes.",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.00019198144,
        0.049884938,
        -0.04002886,
        0.060370345,
        0.033411548,
        -0.014912473,
        -0.046218947,
        -0.028574374,
        0.040303033,
        -0.20336324,
        -0.024940386,
        -0.00475318,
        0.06619228,
        -0.03021024,
        -0.034784455,
        -0.04700491,
        0.09033008,
        -0.03152764,
        -0.020086266,
        -0.10868871,
        -0.03591822,
        -0.09715007,
        -0.034440137,
        -0.09841719,
        0.005941104,
        0.09002172,
        -0.047055576,
        0.02255031,
        -0.09004769,
        0.46793196,
        -0.025566982,
        0.041421227,
        0.051456258,
        -0.078347705,
        0.040634178,
        -0.048466507,
        -0.050807953,
        -0.0061369836,
        -0.025331458,
        0.07842835,
        -0.018265238,
        0.045136046,
        -0.101039,
        -0.015696708,
        -0.1025407,
        -0.026856422,
        -0.09424671,
        0.015488347,
        -0.015485397,
        -0.008630383,
        -0.059304785,
        0.0013214106,
        0.063635714,
        -0.047381226,
        -0.024667004,
        0.004687959,
        -0.0369623,
        -0.025585208,
        0.0081743235,
        -0.01503034,
        -0.06641323,
        0.043647703,
        -0.07481438,
        -0.04772914,
        -0.044513293,
        -0.050481446,
        0.10720931,
        -0.033028983,
        0.025828192,
        0.043170795,
        -0.0039405143,
        -0.03745643,
        -0.028708799,
        0.020641204,
        -0.08349422,
        -0.08763109,
        -0.059434664,
        0.053965013,
        -0.02252259,
        0.0075971084,
        -0.037518777,
        0.07604093,
        -0.075999394,
        -0.018540079,
        -0.005838263,
        -0.03948262,
        -0.033516347,
        0.0410988,
        -0.048577275,
        0.05584443,
        0.0037350776,
        -0.022369606,
        -0.02948487,
        -0.03521904,
        -0.027541833,
        0.040752497,
        -0.11331344,
        0.07925709,
        -0.09592554,
        0.037012532,
        -0.0687062,
        -0.019272376,
        0.010970642,
        -0.017519338,
        0.011840984,
        -0.045013018,
        -0.03510784,
        -0.006927015,
        0.0041904906,
        -0.037644327,
        0.013869436,
        -0.6275355,
        -0.0300985,
        -0.013835853,
        -0.013929413,
        -0.17863151,
        0.07614871,
        -0.006897946,
        0.101161085,
        0.08726402,
        0.0028293033,
        0.05238446,
        -0.011942055,
        -0.007813071,
        -0.05053232,
        0.00812475,
        0.030805865,
        0.0005597343
      ]
    },
    {
      "id": "worldsfair-speaker-227",
      "conference": "worldsfair",
      "name": "Tomás Hernando Kofman",
      "role": "",
      "company": "Notdiamond",
      "twitter": "",
      "talks": [
        {
          "title": "Intelligent Model Routing: Frontier Performance Without Frontier Bills",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.017792534,
        0.050213378,
        -0.0059547583,
        0.019484058,
        0.011423338,
        -0.0106195295,
        -0.056582455,
        0.09247843,
        0.07026558,
        -0.21141176,
        -0.04276443,
        -0.027900474,
        -0.03552382,
        0.01203475,
        -0.0824828,
        0.0142134195,
        0.10697609,
        -0.059359264,
        0.048573766,
        -0.061892133,
        -0.10854549,
        -0.06317089,
        -0.043265786,
        0.06536889,
        -0.018014252,
        0.04835456,
        0.037638184,
        -0.027663248,
        -0.11056924,
        0.50226283,
        -0.11829714,
        0.018313102,
        0.029742057,
        -0.015826348,
        0.048869975,
        -0.061762895,
        -0.048971217,
        0.0068060346,
        -0.01894247,
        0.023509016,
        -0.04319876,
        0.039905768,
        -0.04058272,
        -0.0006402376,
        -0.10342825,
        -0.019200642,
        -0.037740707,
        -0.026549276,
        -0.022247624,
        -0.023629194,
        -0.16986497,
        0.026491573,
        0.005093174,
        0.048342127,
        0.056573756,
        -0.055726502,
        0.058978293,
        -0.05841301,
        0.12795326,
        -0.057318494,
        -0.06912055,
        -0.017086374,
        -0.01972975,
        0.017476808,
        0.011482172,
        -0.02827482,
        0.035812087,
        -0.011354885,
        -0.021193288,
        -0.021178005,
        0.016201809,
        -0.060749702,
        -0.06880668,
        0.0014106557,
        0.0045162076,
        -0.038980197,
        -0.01377406,
        0.124492064,
        0.0022013728,
        -0.008314047,
        0.0072468617,
        0.01685836,
        -0.04411581,
        -0.031810146,
        0.06119112,
        -0.042827897,
        -0.046610966,
        -0.009509272,
        0.053080533,
        0.07082609,
        0.044222895,
        -0.004117762,
        -0.019316938,
        -0.035807323,
        -0.034963302,
        -0.04516549,
        -0.002297796,
        0.055387676,
        0.04184958,
        -0.052497733,
        0.023798171,
        -0.024193618,
        0.06117613,
        -0.024057543,
        -0.10307786,
        -0.036128797,
        -0.023587635,
        -0.04907745,
        0.06784721,
        -0.016179629,
        -0.022790033,
        -0.60159194,
        0.044188704,
        0.000115614,
        -0.040900458,
        -0.15646508,
        0.0057267216,
        0.031049693,
        0.034395434,
        0.062386565,
        0.04178443,
        0.030396372,
        0.01872718,
        0.02874661,
        0.045879845,
        0.055217735,
        0.048955366,
        0.029239442
      ]
    },
    {
      "id": "worldsfair-speaker-228",
      "conference": "worldsfair",
      "name": "Dylan Couzon",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Frontier Is Coming Home",
          "description": "In 2022, the smallest model to clear 60 percent on MMLU had 540 billion parameters. Two years later a 3.8 billion parameter model did the same thing, small enough to run on a phone. That is a 142x drop to reach the same capability floor, and it is the cleanest way to see a trend most people are not pricing in. Call it the lag: the time between a capability showing up at the frontier and that capability running on hardware you own. Today the lag is measured in months, and it keeps shrinking. But raw capability is only half of what makes a model useful. A model that can reason but cannot remember is a stranger every time you talk to it. The other half of local AI is memory, and that half is already here. On-device retrieval has been ready to run locally longer than the models have. The embedding models that power it are tiny, the indexes fit in memory, and none of it touches a network. When your reasoning and your memory both live on your machine, so does your context. Your history, your documents, your past conversations never leave the device. That is the part of this shift that matters most, and the part people overlook because they are busy watching the models. The same shift flips the economics. At 200 dollars a month per seat, a local machine starts to pay for itself in under two years, and the frontier labs' own published usage numbers put heavy coding in the same range. I'll walk through the math, the hardware, and where local still loses. None of this is a bet against scale, or against the Bitter Lesson. The frontier still grows in the data center. The point is that a usable copy keeps arriving on your desk, on a lag, with a memory of its own, for close to free.",
          "day": "Day 3 — Session Day 2",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04211352,
        0.0026015511,
        -0.049056213,
        0.032370165,
        0.07569463,
        -0.052443434,
        -0.068862215,
        -0.021026783,
        0.058714774,
        -0.2257373,
        0.042884655,
        -0.0339876,
        -0.025508521,
        0.0022779172,
        -0.12508257,
        0.060919657,
        0.11264156,
        -0.06545008,
        -0.062326062,
        -0.042749688,
        -0.068598755,
        -0.16344956,
        -0.043740872,
        0.051301368,
        -0.015546728,
        0.106233776,
        -0.0472665,
        0.047392245,
        -0.11924199,
        0.36255792,
        -0.097859815,
        -0.015522883,
        0.03699368,
        -0.030653093,
        0.029676763,
        -0.073219694,
        0.028463742,
        0.04173982,
        0.056559686,
        0.05312909,
        0.03533221,
        0.014544769,
        -0.051080365,
        0.015985465,
        -0.042556424,
        0.0063025076,
        -0.08471786,
        -0.020284835,
        -0.021039793,
        0.0016734952,
        -0.064100765,
        0.014839344,
        0.043451015,
        0.07076028,
        0.05319846,
        0.057535227,
        -0.0025058105,
        -0.054600228,
        0.15076497,
        -0.051962133,
        -0.06402622,
        -0.0239645,
        -0.012766811,
        0.055590115,
        -0.013900081,
        0.012748427,
        -0.01995077,
        -0.071217075,
        -0.026994685,
        -0.056773685,
        0.09551374,
        0.016367577,
        -0.042059656,
        -0.005755447,
        -0.08963789,
        -0.0951625,
        -0.038709447,
        0.056021284,
        -0.010967856,
        0.011559321,
        0.009877399,
        0.016574942,
        -0.05474672,
        -0.010405641,
        0.09232355,
        -0.016645575,
        -0.06961716,
        0.072808675,
        0.03310346,
        0.07974012,
        0.0671029,
        -0.015880166,
        -0.022271328,
        0.026965393,
        -0.038768742,
        0.006005859,
        0.0070508043,
        0.005366279,
        -0.02161975,
        -0.004829433,
        0.07066036,
        -0.07541265,
        -0.004990319,
        -0.06963391,
        -0.030258762,
        -0.09529167,
        0.0042312522,
        0.03718036,
        -0.015956543,
        0.009463439,
        -0.017600728,
        -0.6637316,
        0.004793712,
        0.046125878,
        -0.027376058,
        -0.14978741,
        0.019877074,
        0.035221856,
        0.07255615,
        0.064954706,
        0.0029695558,
        0.036887042,
        -0.0031764081,
        0.028743198,
        -0.0034305647,
        -0.010081475,
        0.025707757,
        0.0025174774
      ]
    },
    {
      "id": "worldsfair-speaker-229",
      "conference": "worldsfair",
      "name": "Roland Gavrilescu",
      "role": "Founder",
      "company": "Introspection",
      "twitter": "https://x.com/rolandgvc",
      "talks": [
        {
          "title": "Autoresearch in the wild",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06386054,
        0.0058288956,
        0.047812916,
        -0.012964809,
        0.07289402,
        -0.012021079,
        0.013736833,
        0.03303925,
        0.08197329,
        -0.2280916,
        -0.045710914,
        0.013719143,
        0.041439638,
        -0.032420874,
        -0.039905366,
        0.03514792,
        0.06808351,
        -0.039246194,
        -0.049923517,
        -0.15251556,
        -0.060903527,
        -0.10536293,
        -0.08399578,
        0.0666954,
        -0.0024476608,
        0.107936986,
        -0.019698812,
        -0.0052065933,
        -0.13991433,
        0.5043876,
        -0.03490306,
        -0.018484581,
        0.021256085,
        -0.023791421,
        0.054995418,
        -0.058256343,
        -0.004787303,
        0.030928804,
        -0.025691256,
        0.07244135,
        -0.068801165,
        0.01646063,
        0.018686384,
        0.037000973,
        0.05893086,
        0.009547842,
        -0.07341947,
        -0.056952875,
        -0.015055626,
        -0.013115148,
        -0.05181045,
        0.02177212,
        0.07610122,
        -0.06631156,
        0.034763638,
        0.038450353,
        -0.0871599,
        -0.078704245,
        0.021585234,
        -0.03164536,
        -0.043611933,
        -0.017297737,
        0.008105699,
        0.05244352,
        -0.02819483,
        0.012571713,
        0.16048229,
        -0.050827622,
        0.012519378,
        -0.008175868,
        0.053635508,
        0.010108158,
        0.01922082,
        0.021211905,
        -0.053488858,
        -0.016113903,
        -0.05154193,
        0.06925584,
        0.006838711,
        -0.050188925,
        0.052271225,
        0.037494652,
        -0.04870894,
        -0.04356558,
        -0.03303927,
        0.0013842484,
        -0.06470945,
        0.05489127,
        -0.021678401,
        0.0136632575,
        0.05390944,
        -0.06115412,
        -0.03555457,
        -0.011123271,
        -0.021226458,
        0.016613496,
        -0.0994142,
        0.08554177,
        0.034175497,
        0.008415263,
        -0.07176025,
        -0.031210044,
        0.030637734,
        -0.06601368,
        -0.04447314,
        -0.042224836,
        -0.026657753,
        -0.00999859,
        0.008370638,
        0.0033156537,
        0.01992168,
        -0.5904243,
        0.06353302,
        -0.015596446,
        -0.045658033,
        -0.1725067,
        -0.04367124,
        0.0036240723,
        0.02171887,
        0.039869677,
        0.024503496,
        0.01565041,
        -0.02708674,
        0.043109722,
        0.023131762,
        0.0071243243,
        0.02117743,
        0.060068924
      ]
    },
    {
      "id": "worldsfair-speaker-230",
      "conference": "worldsfair",
      "name": "Julian Bright",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Autoresearch in the wild",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0020958958,
        0.019193262,
        -0.024705423,
        0.041700713,
        0.06842206,
        -0.08958509,
        -0.082753524,
        0.037059736,
        0.06374622,
        -0.21822782,
        -0.04077507,
        0.04802442,
        -0.0051581822,
        0.032897554,
        0.017079046,
        0.06425825,
        0.11707422,
        -0.05613609,
        -0.08210197,
        -0.08360331,
        -0.02473701,
        -0.13972233,
        -0.061575085,
        0.04398996,
        -0.00828697,
        0.11491862,
        -0.005737616,
        0.014420993,
        -0.09997629,
        0.39743578,
        -0.10112985,
        -0.019868098,
        0.064803354,
        -0.002319745,
        0.027988542,
        -0.01673893,
        0.008954279,
        0.0012102107,
        0.035283457,
        0.09666303,
        -0.06776002,
        0.0043177493,
        -0.0029199237,
        0.10183215,
        0.09907385,
        0.033680182,
        -0.09206712,
        -0.047600236,
        -0.052519858,
        -0.0077841277,
        -0.10235758,
        -0.0075497148,
        -0.009998536,
        -0.04036588,
        -0.026390497,
        -0.0037294927,
        -0.042826876,
        -0.089175604,
        0.059228677,
        -0.047001414,
        -0.03261584,
        -0.07190171,
        -0.02693211,
        0.0038544163,
        -0.0073291888,
        0.03275337,
        0.12588948,
        -0.03761038,
        0.0124293165,
        -0.018488301,
        -0.05500116,
        0.026563017,
        -0.006229794,
        0.0043890136,
        -0.113687254,
        -0.00838262,
        -0.06258242,
        0.074562915,
        -0.012011181,
        -0.020649597,
        0.029589914,
        0.07947062,
        0.003836467,
        0.009983413,
        -0.068208896,
        0.0007615835,
        -0.07351852,
        0.06781984,
        -0.0497914,
        0.06152075,
        0.02444577,
        -0.050497822,
        -0.03748547,
        0.009303959,
        0.07426834,
        0.04526977,
        -0.05841573,
        0.11797558,
        0.12503032,
        -0.0081158625,
        -0.0640633,
        -0.030830877,
        0.07323082,
        -0.07397503,
        -0.009299519,
        -0.07563407,
        -0.020269806,
        -0.020664936,
        0.03239715,
        -0.08502233,
        -0.006301549,
        -0.6032771,
        -0.006482558,
        -0.10167453,
        -0.040216755,
        -0.14495218,
        0.01789405,
        0.058341704,
        0.013625862,
        0.0821176,
        -0.0034524484,
        0.030978665,
        0.040828664,
        -0.034048807,
        -0.08739375,
        0.075535074,
        0.013944614,
        -0.027709236
      ]
    },
    {
      "id": "worldsfair-speaker-231",
      "conference": "worldsfair",
      "name": "Robert Brennan",
      "role": "",
      "company": "OpenHands",
      "twitter": "https://x.com/rbren_dev",
      "talks": [
        {
          "title": "Sandboxes Aren't Optional: Runtime Isolation Patterns for Coding Agents at Scale",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.025952814,
        -0.0010690031,
        -0.0050325026,
        0.060448952,
        0.047433138,
        -0.0037175396,
        -0.093638904,
        0.03377123,
        0.02651167,
        -0.24261758,
        -0.025236271,
        -0.052845467,
        0.0048687607,
        0.011849551,
        -0.062098388,
        0.03192051,
        0.057884447,
        -0.089741655,
        -0.050462022,
        -0.090300456,
        -0.055999517,
        -0.117176,
        0.027591905,
        0.034956362,
        -0.026329085,
        0.013288812,
        -0.034241453,
        0.10987772,
        -0.10770936,
        0.5016706,
        -0.05753301,
        0.037953325,
        0.033850092,
        -0.013329023,
        0.046308942,
        -0.09712934,
        -0.0068881307,
        0.017545203,
        0.012005408,
        0.030822068,
        0.017735826,
        -0.03081175,
        -0.027114486,
        0.0103968335,
        -0.042506896,
        0.04624278,
        -0.03136382,
        -0.05253234,
        0.01875594,
        -0.07443905,
        -0.024446487,
        0.09894433,
        0.034847155,
        0.0118442485,
        0.06877221,
        -0.043517254,
        -0.020270966,
        -0.041245457,
        0.06713213,
        -0.007974147,
        -0.042330217,
        0.012729736,
        -0.044431157,
        -0.030705178,
        -0.06296126,
        -0.033958048,
        0.11348542,
        -0.08017499,
        0.008749502,
        -0.00434083,
        0.036353778,
        -0.0118189035,
        -0.058057643,
        0.027693756,
        -0.07875048,
        -0.038288042,
        -0.033284437,
        0.015447419,
        -0.038035918,
        -0.011205621,
        0.029185986,
        0.1338094,
        -0.02829523,
        -0.04047425,
        0.015229301,
        -0.10993805,
        -0.055142105,
        0.06967409,
        0.065699555,
        0.060712513,
        0.06694709,
        -0.0029932056,
        0.016549315,
        0.044940587,
        -0.026990153,
        -0.016693087,
        -0.021263843,
        0.062506765,
        -0.008842537,
        0.0241207,
        -0.06577881,
        -0.003793403,
        0.030901076,
        -0.05056648,
        -0.038280144,
        -0.028623814,
        -0.0044669416,
        -0.06714653,
        0.10030775,
        -0.043019824,
        -0.0051899627,
        -0.5979954,
        -0.07459603,
        -0.06295213,
        -0.024028556,
        -0.14806587,
        0.044537432,
        0.0066663017,
        -0.0054279156,
        0.03908588,
        -0.04211142,
        0.048338704,
        -0.02109083,
        -0.015054585,
        0.011545932,
        0.010805861,
        0.06764111,
        -0.0057332627
      ]
    },
    {
      "id": "worldsfair-speaker-232",
      "conference": "worldsfair",
      "name": "Amit Navindgi",
      "role": "",
      "company": "Zoox",
      "twitter": "https://x.com/amitnavindgi",
      "talks": [
        {
          "title": "From Self-Driving Monorepo to Self-Driving Cars",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.038232636,
        -0.0030637628,
        -0.017451731,
        0.080544055,
        0.0048371986,
        0.008906892,
        -0.048153833,
        -0.045491338,
        0.07606623,
        -0.2267542,
        0.0077880486,
        -0.041567113,
        0.041983433,
        -0.051833525,
        -0.037418727,
        -0.004386994,
        0.0641937,
        -0.03322624,
        -0.047129393,
        -0.07472605,
        -0.06709484,
        -0.0679239,
        0.015725013,
        -0.031752076,
        0.001139723,
        0.054141007,
        0.005427961,
        0.0005899299,
        -0.124216095,
        0.5212031,
        -0.11670089,
        -0.020609928,
        -0.009687915,
        -0.043725707,
        0.03701555,
        -0.059817426,
        -0.012077454,
        -0.015154204,
        -0.0060946546,
        0.000088488174,
        0.0006928371,
        0.030661616,
        -0.07113201,
        -0.008822158,
        0.021657255,
        0.0006650972,
        -0.10753542,
        -0.04516218,
        -0.002528658,
        -0.042285476,
        -0.050934784,
        -0.036627628,
        0.013565882,
        -0.057243705,
        -0.0072422363,
        0.07287131,
        -0.027017282,
        -0.10292418,
        0.039508045,
        -0.034178186,
        -0.08478391,
        -0.031296335,
        0.029589277,
        -0.028910829,
        -0.008497984,
        0.0069071907,
        0.06995781,
        -0.09650506,
        -0.028328884,
        0.055445623,
        0.04325819,
        0.009557227,
        -0.046990886,
        0.030704653,
        -0.06296638,
        -0.016561147,
        -0.0010610028,
        0.09122914,
        -0.049597334,
        0.008050536,
        0.03715836,
        -0.019904584,
        -0.10152132,
        -0.015220399,
        -0.0006491174,
        -0.019236937,
        -0.021225948,
        0.04346476,
        -0.017557248,
        0.039872974,
        0.06999947,
        0.008370674,
        0.017911002,
        0.008251386,
        -0.07083555,
        0.06889985,
        -0.09076863,
        0.06628036,
        -0.022478903,
        0.03702624,
        -0.013756624,
        -0.05423609,
        0.014133088,
        0.040426556,
        -0.011997579,
        -0.08401769,
        0.025888659,
        -0.027732283,
        0.030598773,
        -0.039752964,
        -0.07190564,
        -0.61012614,
        -0.04079568,
        -0.0193238,
        -0.0016276459,
        -0.16751538,
        -0.019667309,
        -0.03989203,
        0.026457602,
        0.096065156,
        0.0074617486,
        0.064472854,
        -0.03397749,
        -0.016665362,
        -0.003320514,
        0.006176415,
        -0.040785305,
        -0.031910673
      ]
    },
    {
      "id": "worldsfair-speaker-233",
      "conference": "worldsfair",
      "name": "Shlok Khemani",
      "role": "Writer & Programmer",
      "company": "Independent",
      "twitter": "https://x.com/shloked",
      "talks": [
        {
          "title": "Lessons from Studying Every Memory System",
          "description": "I've studied every major memory implementation in the industry and then built multiple memory systems for various teams. Sharing hot takes and lessons from across architectures, design, scaling, evals, and memory philosophy.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.024656499,
        0.099824086,
        0.0130398115,
        -0.003861654,
        0.029733015,
        -0.06360027,
        -0.062372882,
        0.012116616,
        0.12546013,
        -0.230714,
        -0.023702217,
        -0.092957646,
        -0.010641494,
        -0.09957698,
        -0.10490931,
        -0.035222434,
        0.031299572,
        -0.0650035,
        -0.011908632,
        -0.10223075,
        0.019490873,
        -0.0522704,
        -0.046638798,
        0.09068287,
        -0.09651692,
        0.069587044,
        -0.007046125,
        0.007456663,
        -0.20005904,
        0.46298018,
        -0.06595508,
        0.029232435,
        0.023381662,
        -0.07551579,
        0.11811118,
        -0.042720858,
        0.046037413,
        0.038740683,
        0.009741247,
        0.043010447,
        -0.0357064,
        0.021945495,
        -0.0014441763,
        0.00430767,
        -0.090396754,
        -0.043910217,
        -0.038428094,
        0.022410512,
        0.004254986,
        -0.03451011,
        -0.0659848,
        -0.0121725295,
        0.106557,
        0.0075532543,
        -0.051106226,
        -0.0068593076,
        0.06974644,
        -0.070116475,
        0.06142464,
        -0.044220686,
        -0.081287496,
        0.0023115799,
        0.029693859,
        0.0033986836,
        0.04090024,
        -0.08634144,
        0.07916772,
        0.02420242,
        -0.056275573,
        0.026030546,
        0.015107819,
        0.015739182,
        -0.03647528,
        0.044471864,
        -0.08454001,
        -0.12235888,
        -0.042159334,
        0.07615643,
        0.008926524,
        -0.011538335,
        -0.005114143,
        0.0280249,
        0.01576311,
        -0.063097544,
        0.017750368,
        -0.009584473,
        -0.041782845,
        0.07558989,
        -0.038807612,
        0.03952388,
        0.06315517,
        -0.029353756,
        -0.0019662732,
        -0.004068692,
        -0.035687644,
        0.020580946,
        0.011968267,
        0.12416442,
        -0.024363102,
        0.012141315,
        0.003817382,
        0.011592274,
        0.030667411,
        -0.03360752,
        0.0057055308,
        0.012167766,
        0.03138475,
        -0.0010423434,
        0.035052184,
        -0.014076304,
        -0.017569872,
        -0.6016511,
        -0.03718924,
        0.031970292,
        -0.034603667,
        -0.12716609,
        0.015038328,
        -0.029076712,
        -0.01002696,
        0.035129827,
        0.0032086624,
        0.019746795,
        -0.040066183,
        0.0028222126,
        -0.05221777,
        0.071767025,
        0.029917674,
        -0.00096848013
      ]
    },
    {
      "id": "worldsfair-speaker-234",
      "conference": "worldsfair",
      "name": "Ishan Anand",
      "role": "Chief AI Officer (CAIO)",
      "company": "InsightSciences.ai",
      "twitter": "",
      "talks": [
        {
          "title": "Will AI predict people like we predict the weather? (alternate title “A field guide to synthetic personas for market research”)",
          "description": "Large language models can now stand in for humans in surprising ways, from predicting personality types to replicating their responses in market research. Like weather forecasting, once considered impossible and now so routine we take it for granted, LLMs are in the early, unreliable-but-improving stage of simulating how populations think and respond. Teams are already using LLMs as synthetic survey respondents for concept testing, UX exploration, and early market validation. In the past year, the field has gotten both more promising and more tricky. The real question is no longer \"can LLMs simulate people?\", but whether the simulation is validated for the decision you want to make. New methods show that how you ask an LLM matters as much as which model you use and can dramatically improve fidelity to real human responses. Meanwhile validation studies show accuracy can mask subgroup distortion and that seemingly minor choices can reshape the simulated population entirely. This talk gives entrepreneurs, engineers, and PMs an overview of the techniques and a framework for validating synthetic respondents before making decisions. Even if you never build a synthetic persona, this is one of the richest windows into LLM behavior under the hood and these lessons apply to any system where you're trusting an LLM to represent something about the real world.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.053302627,
        0.09102928,
        -0.014984031,
        0.025471043,
        0.013377852,
        -0.025443392,
        -0.007421169,
        -0.01515145,
        0.07726247,
        -0.2571976,
        -0.0041788374,
        -0.04387593,
        -0.0010614302,
        -0.061675124,
        -0.07768721,
        0.018555144,
        0.07455016,
        -0.0058448873,
        -0.05107616,
        -0.064976506,
        -0.023555377,
        -0.06399286,
        0.015505392,
        0.122672476,
        -0.046448197,
        0.10539083,
        -0.022338474,
        0.05363099,
        -0.10876662,
        0.42805934,
        -0.090466894,
        -0.016988339,
        0.011387172,
        -0.042395417,
        0.013526112,
        -0.02963246,
        -0.038867276,
        -0.03305999,
        0.036298722,
        0.005023459,
        0.011507298,
        0.032219708,
        0.00037032593,
        0.048104312,
        0.03848223,
        0.0151511105,
        -0.024189107,
        0.005876774,
        0.022330873,
        -0.05198592,
        -0.0847302,
        0.0015275137,
        0.08359276,
        -0.023794712,
        0.023434523,
        0.056184795,
        0.02332133,
        -0.11999543,
        0.06803621,
        -0.05709401,
        -0.134343,
        -0.03790367,
        -0.021379342,
        0.0030413617,
        0.048807103,
        -0.03933338,
        0.06958105,
        -0.018925406,
        -0.05757377,
        -0.021871256,
        0.004970525,
        0.06294441,
        -0.054984313,
        -0.038105544,
        -0.05958035,
        -0.071689405,
        -0.09631182,
        0.13246274,
        -0.011411151,
        0.022720473,
        0.017858423,
        -0.01104487,
        -0.07322837,
        0.045963388,
        -0.013749616,
        -0.049232177,
        -0.009161926,
        0.016587606,
        -0.003961637,
        0.07828442,
        0.047782905,
        -0.012226343,
        -0.04505237,
        -0.00014771428,
        -0.07542371,
        0.028139472,
        -0.0717225,
        0.031024978,
        0.026260799,
        0.017831089,
        0.021337345,
        0.019980323,
        0.12584233,
        0.024451163,
        -0.07493104,
        -0.032305855,
        0.08420085,
        -0.0063248365,
        0.035311874,
        -0.028286211,
        -0.02500792,
        -0.6221292,
        0.084718004,
        0.027481344,
        -0.0019266669,
        -0.13115925,
        0.11337514,
        -0.019921618,
        0.0068003754,
        0.03430783,
        -0.011803599,
        0.052008163,
        -0.033252295,
        -0.018226823,
        -0.085686915,
        0.052130386,
        0.062462483,
        0.040564455
      ]
    },
    {
      "id": "worldsfair-speaker-235",
      "conference": "worldsfair",
      "name": "Jeremiah Lowin",
      "role": "",
      "company": "Prefect",
      "twitter": "",
      "talks": [
        {
          "title": "Generative UI... in Python?",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.07241968,
        -0.018238027,
        -0.032546997,
        0.018464237,
        0.021629388,
        -0.030290056,
        -0.0978797,
        0.017463632,
        0.07650581,
        -0.1876412,
        -0.048251927,
        -0.05315827,
        0.047601726,
        -0.020273972,
        -0.03230455,
        -0.011221116,
        0.09873415,
        -0.10362605,
        -0.037037093,
        -0.089246385,
        -0.07655482,
        -0.110137045,
        0.017899446,
        0.03089236,
        -0.04965688,
        0.11401376,
        0.0025400957,
        0.035326358,
        -0.11423974,
        0.4963307,
        -0.07541192,
        0.032151718,
        0.023522614,
        -0.021316051,
        -0.02759424,
        -0.07796773,
        -0.0237885,
        -0.024484565,
        0.01893683,
        0.102862,
        0.025947008,
        -0.0038186018,
        -0.010911637,
        -0.083951905,
        -0.019627573,
        -0.037836425,
        -0.08067598,
        -0.036391735,
        0.050037373,
        -0.018453209,
        -0.084400184,
        0.04176006,
        0.013108937,
        -0.018331049,
        0.061966576,
        0.0007251938,
        -0.019459821,
        -0.0843965,
        0.0818754,
        -0.07727409,
        -0.057515502,
        -0.00079200696,
        -0.025850851,
        -0.016038544,
        0.024265891,
        -0.06476929,
        0.05137153,
        -0.019355552,
        0.05421016,
        0.0016227096,
        0.0020398966,
        0.0216662,
        -0.04045504,
        0.018328203,
        -0.04316031,
        -0.018517697,
        0.00648116,
        0.1125248,
        0.013103851,
        0.04314717,
        0.00797464,
        -0.025442572,
        -0.08333059,
        -0.06367432,
        -0.033012856,
        -0.052782927,
        0.0015678592,
        0.037351917,
        -0.04524683,
        0.1059243,
        0.0020219658,
        -0.04005861,
        -0.023285415,
        -0.008522656,
        -0.036936127,
        -0.026509827,
        -0.03210413,
        0.02765104,
        0.031136423,
        0.011473983,
        -0.09310678,
        -0.03197741,
        0.044308323,
        -0.04724675,
        -0.07510649,
        -0.062034998,
        0.025856484,
        -0.009694219,
        0.0015429844,
        0.0043488997,
        0.0051065865,
        -0.6136013,
        -0.05157224,
        0.006947227,
        0.068201564,
        -0.14640357,
        0.0863937,
        0.016898308,
        0.025508257,
        0.053729884,
        0.01826044,
        0.047214247,
        0.013917711,
        0.007084327,
        0.028176012,
        0.08400087,
        0.05172006,
        -0.012957155
      ]
    },
    {
      "id": "worldsfair-speaker-236",
      "conference": "worldsfair",
      "name": "Patricija Žemaitytė",
      "role": "Product Manager",
      "company": "Oxylabs",
      "twitter": "",
      "talks": [
        {
          "title": "How Web Data Infrastructure Powers the Next Generation of AI",
          "description": "For years, the web intelligence industry has powered major data developments. As big data grew, ensuring sustained data flow became harder. Now, AI is taking the biggest leaps forward. How the web intelligence industry responded to this increasing scale and complexity is the story of the most crucial steps forward in AI today.\n\nThis presentation demonstrates how web scraping infrastructure fuels AI innovation by linking the web's repository to AI developers. Told through AI products, it addresses both the engineering challenges and solutions for developers, and the strategic use cases for business decision-makers.\n\nSummary: How web scraping infrastructure drives AI innovation by solving engineering challenges and enabling strategic business use cases.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0366409,
        0.023702515,
        0.004734037,
        -0.022950938,
        -0.016057918,
        -0.04876972,
        -0.11269633,
        -0.026678944,
        0.021778733,
        -0.20502423,
        -0.032085545,
        -0.034300227,
        -0.02286612,
        -0.028089782,
        -0.049721677,
        0.015572405,
        0.09599556,
        -0.05717539,
        0.0022232917,
        -0.071602754,
        -0.018457847,
        -0.061181724,
        0.009529309,
        0.07290418,
        -0.042566873,
        0.098778106,
        -0.009830716,
        0.016495738,
        -0.09987154,
        0.50772977,
        -0.12116733,
        0.016165234,
        0.006533196,
        0.02037665,
        0.048631787,
        -0.013564641,
        -0.04283834,
        -0.035946704,
        -0.013191802,
        0.08088382,
        -0.0042115785,
        0.10336692,
        -0.029127005,
        0.030138483,
        -0.013662135,
        0.02704213,
        -0.074384615,
        -0.015461481,
        0.006974346,
        -0.024267247,
        -0.026339086,
        0.039085776,
        0.035453048,
        -0.02185272,
        0.050534435,
        0.043078333,
        -0.004917433,
        -0.08201873,
        0.065896355,
        -0.060225137,
        -0.099773295,
        -0.03174732,
        0.028149286,
        0.041044597,
        0.059660286,
        -0.08196971,
        0.039601542,
        -0.10726112,
        -0.021365134,
        -0.022402968,
        -0.04035255,
        -0.038289264,
        -0.025989035,
        0.048247628,
        -0.04928995,
        -0.07108191,
        -0.053801186,
        0.049610823,
        -0.023163112,
        0.054413393,
        0.009180067,
        0.06579925,
        0.038109913,
        -0.001043027,
        0.073785976,
        0.028847307,
        -0.07084444,
        -0.0049155485,
        -0.019294392,
        0.07143397,
        -0.0024076258,
        -0.023557702,
        -0.09373188,
        0.0045919516,
        -0.06943571,
        0.00544501,
        -0.023697723,
        0.0785502,
        0.047648877,
        0.053880315,
        -0.05904402,
        -0.028107885,
        -0.001877048,
        0.022665767,
        -0.00043124327,
        -0.06616556,
        0.02169768,
        -0.040708274,
        0.085630015,
        0.052576005,
        -0.028925207,
        -0.6040877,
        0.041715555,
        0.0132510755,
        0.032076348,
        -0.1694223,
        -0.02094224,
        0.0017666741,
        0.024515651,
        0.091072224,
        0.006169495,
        0.07238465,
        -0.055556886,
        -0.0025831768,
        0.05741691,
        0.04848276,
        0.018558541,
        0.061736695
      ]
    },
    {
      "id": "worldsfair-speaker-237",
      "conference": "worldsfair",
      "name": "Alex Hancock",
      "role": "AI Agent",
      "company": "Block",
      "twitter": "https://x.com/alexjhancock",
      "talks": [
        {
          "title": "The Universal Remote Control for AI",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.039354168,
        0.02308275,
        0.0078071533,
        -0.027158532,
        0.030132256,
        0.0078031016,
        -0.06537755,
        -0.007869611,
        0.02413754,
        -0.24714069,
        0.02069147,
        -0.07941047,
        -0.025972888,
        -0.018969657,
        0.0016604845,
        0.0052213785,
        0.05280988,
        -0.10530612,
        0.029517962,
        -0.1062523,
        -0.08110342,
        -0.07402028,
        -0.05507721,
        0.01269556,
        -0.099788055,
        0.07024014,
        -0.04134053,
        0.014799599,
        -0.09282207,
        0.43726873,
        -0.026597092,
        0.04171124,
        -0.027482979,
        -0.015281005,
        0.07006635,
        -0.07419766,
        -0.08907283,
        0.0017436926,
        0.029697964,
        0.042150594,
        -0.03771251,
        0.04764443,
        -0.024871092,
        0.034053136,
        -0.037221763,
        -0.010128544,
        -0.081595495,
        -0.024353195,
        -0.025325906,
        -0.013892109,
        -0.10585069,
        0.07119767,
        0.07209525,
        -0.058406357,
        0.004537442,
        0.00055489206,
        -0.02175896,
        -0.101485014,
        0.05399345,
        -0.06410909,
        -0.041154023,
        -0.024069058,
        -0.063146085,
        -0.022669684,
        0.0348558,
        -0.037824336,
        0.06831442,
        -0.0694266,
        -0.016172465,
        -0.08374904,
        -0.033021867,
        0.016700082,
        -0.05579355,
        -0.011433558,
        -0.045026947,
        -0.08324085,
        -0.09097671,
        0.052909177,
        0.021928933,
        -0.015776215,
        -0.034617137,
        0.043794308,
        -0.022905055,
        -0.05022373,
        -0.012284682,
        -0.07053118,
        -0.06239054,
        0.06292275,
        -0.0011647718,
        0.054675035,
        0.053644367,
        0.028552806,
        -0.09701554,
        -0.02386668,
        -0.11288394,
        0.035051305,
        -0.14741369,
        0.041767504,
        0.020875955,
        0.019692518,
        -0.07981955,
        -0.060088255,
        0.06362609,
        -0.04013216,
        -0.0046900488,
        -0.021862436,
        0.017943691,
        -0.044912864,
        0.06613745,
        -0.044173993,
        0.019741513,
        -0.61675555,
        -0.022189863,
        -0.032289624,
        -0.03292266,
        -0.12917411,
        0.04208235,
        0.06491794,
        0.036185205,
        0.061143585,
        0.0072649145,
        0.04811865,
        -0.0783011,
        0.09148516,
        0.04830126,
        0.02153271,
        -0.016959768,
        -0.02695138
      ]
    },
    {
      "id": "worldsfair-speaker-238",
      "conference": "worldsfair",
      "name": "Mahesh Sathiamoorthy",
      "role": "CEO",
      "company": "Bespoke Labs",
      "twitter": "",
      "talks": [
        {
          "title": "HOLD — Bespoke Labs",
          "description": "Hold for Bespoke Labs. Company works on data curation, eval tooling, and reinforcement-learning environment curation for agent development.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.03757756,
        0.08442858,
        0.0722477,
        0.05166333,
        -0.010671685,
        -0.011479376,
        -0.056194995,
        -0.0016450201,
        -0.0011789476,
        -0.24390918,
        -0.00092032464,
        -0.04049644,
        -0.02498505,
        0.014112562,
        -0.05353971,
        -0.044920668,
        0.0105125895,
        -0.10559431,
        -0.06331381,
        -0.019892626,
        -0.0114453565,
        -0.042019937,
        -0.047385003,
        0.083873145,
        0.0077014226,
        0.08421605,
        -0.0076189726,
        0.06281491,
        -0.08377596,
        0.46474674,
        -0.12912035,
        -0.014618466,
        0.0021193891,
        0.05260313,
        0.035416823,
        -0.044895884,
        -0.013884221,
        0.013833863,
        -0.00070618634,
        0.0066433535,
        -0.053030193,
        0.10966295,
        0.037517577,
        0.058214422,
        0.027877992,
        0.019615706,
        0.01935718,
        -0.038519014,
        -0.011470387,
        0.003947126,
        -0.00067095953,
        0.01152785,
        0.05952955,
        0.09397395,
        0.055148255,
        0.015410587,
        0.014357817,
        -0.024820784,
        0.04844676,
        0.05780232,
        -0.1229022,
        0.0057387743,
        0.046267588,
        -0.094267435,
        -0.011233165,
        -0.026779925,
        0.072217,
        -0.000221323,
        0.02475452,
        -0.013558396,
        0.037057716,
        -0.013736534,
        -0.063931726,
        -0.048706137,
        -0.062224377,
        -0.10544304,
        -0.09007706,
        0.08351036,
        0.038290836,
        -0.025120348,
        -0.04037265,
        0.06428595,
        -0.11916585,
        -0.07624187,
        0.0959955,
        -0.006972612,
        -0.01183271,
        -0.0037531024,
        0.0029731276,
        0.032913037,
        -0.007056907,
        -0.016806269,
        -0.061195973,
        -0.005100658,
        -0.096967325,
        0.026524574,
        -0.07332703,
        0.036071673,
        -0.025236988,
        0.0101107145,
        0.017091228,
        -0.014529793,
        0.09865689,
        0.008771101,
        0.051977403,
        0.011073287,
        0.08845239,
        -0.06842658,
        0.08733028,
        -0.013323437,
        0.0016710317,
        -0.6120408,
        -0.018300168,
        -0.03564823,
        0.0027868117,
        -0.13615602,
        0.015951533,
        -0.06364617,
        0.067361,
        0.039188415,
        -0.010752465,
        0.04514911,
        -0.018819312,
        0.00916206,
        -0.049444035,
        -0.0017705982,
        0.08604709,
        -0.041397177
      ]
    },
    {
      "id": "worldsfair-speaker-239",
      "conference": "worldsfair",
      "name": "Carlos Sanchez",
      "role": "",
      "company": "Adobe",
      "twitter": "https://x.com/csanchez",
      "talks": [
        {
          "title": "Agentic Sites: Building Hyper Personalized Websites",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0078134565,
        0.01997403,
        -0.007930041,
        0.0059338156,
        0.009797878,
        -0.034061044,
        -0.07801218,
        -0.034278154,
        0.0011795178,
        -0.21150623,
        -0.027993545,
        -0.08298532,
        -0.004621212,
        -0.017764755,
        -0.043634266,
        0.042640463,
        0.07729408,
        -0.06939173,
        0.017851925,
        -0.04813527,
        -0.11301015,
        -0.1026272,
        -0.031503655,
        0.010454032,
        -0.06748896,
        0.10379779,
        -0.00008244102,
        -0.023430372,
        -0.16261132,
        0.4861303,
        0.03425154,
        0.0070368745,
        -0.016819268,
        -0.11415083,
        -0.0111522805,
        0.028451126,
        -0.00041135488,
        -0.016906973,
        0.095290765,
        -0.01806682,
        -0.049628712,
        0.034306966,
        -0.083622344,
        0.007718003,
        0.004454729,
        -0.036458343,
        0.0050353324,
        -0.049852185,
        -0.10000237,
        0.019244675,
        -0.11917008,
        0.022103323,
        -0.0038975326,
        -0.01756524,
        -0.042302176,
        -0.03172835,
        -0.0036986675,
        -0.028562991,
        0.0657113,
        -0.13931656,
        -0.100846626,
        -0.048824627,
        0.003350144,
        -0.03422285,
        0.031109376,
        -0.04449828,
        -0.008377146,
        -0.030010333,
        0.030629026,
        -0.047111318,
        -0.031713665,
        0.010740443,
        -0.10648264,
        -0.06886988,
        -0.10557259,
        -0.06052738,
        -0.07293348,
        0.051347777,
        -0.006293582,
        0.07042916,
        -0.022653544,
        0.008482395,
        -0.10451433,
        -0.029161237,
        -0.024121339,
        -0.02418761,
        -0.0583098,
        0.040860336,
        -0.06300021,
        0.016824065,
        0.033410974,
        0.05760509,
        -0.0013130844,
        0.0062475246,
        -0.01077103,
        -0.00021398997,
        -0.05712531,
        0.049355738,
        0.045802128,
        0.0045479103,
        -0.051922068,
        -0.021183198,
        -0.022301365,
        -0.034175314,
        -0.08464459,
        -0.05457907,
        0.0573399,
        0.024736002,
        0.09482455,
        -0.04285715,
        0.025190903,
        -0.5979757,
        -0.08097555,
        -0.0076181837,
        0.025672724,
        -0.100200385,
        0.026031123,
        -0.024979843,
        0.04669962,
        0.052453917,
        0.028073885,
        0.047422186,
        0.05071671,
        -0.048788596,
        0.0009937345,
        0.05509624,
        0.028877702,
        0.034601655
      ]
    },
    {
      "id": "worldsfair-speaker-240",
      "conference": "worldsfair",
      "name": "Kyle Corbitt",
      "role": "",
      "company": "Coreweave",
      "twitter": "",
      "talks": [
        {
          "title": "Inference is the New Training Loop: Architecting High-Reliability Agents and Continuous AI Systems",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.008131032,
        0.04538011,
        0.051755585,
        0.07102861,
        0.020606961,
        -0.025869925,
        -0.021960381,
        -0.031282794,
        0.054304022,
        -0.21543542,
        -0.016647026,
        -0.05257373,
        0.0152596645,
        0.0071261483,
        -0.016409911,
        0.028277827,
        0.109537974,
        -0.102734245,
        -0.023876136,
        -0.09615396,
        -0.07969663,
        -0.07223439,
        0.0029564782,
        0.011882237,
        -0.045911033,
        0.09554218,
        -0.032734666,
        0.091240674,
        -0.10952559,
        0.48058948,
        -0.10601623,
        -0.0025195878,
        0.031344045,
        -0.021407846,
        0.075273715,
        -0.058195844,
        -0.030912641,
        0.041266903,
        0.040947475,
        0.026648184,
        0.010481415,
        -0.011416803,
        -0.077898234,
        0.0031388453,
        -0.05356664,
        0.05509582,
        -0.076463975,
        -0.039591018,
        -0.010821245,
        -0.061825424,
        -0.05428555,
        0.03435835,
        0.03892462,
        -0.019636553,
        0.056072734,
        0.0403707,
        -0.049943957,
        -0.058498263,
        0.07035547,
        -0.008607004,
        -0.1043406,
        -0.018335469,
        -0.03622498,
        0.037257537,
        0.004691156,
        -0.089097865,
        0.06639044,
        -0.024717767,
        -0.025186012,
        -0.030632041,
        -0.02128744,
        0.041756194,
        -0.046970457,
        -0.029933464,
        -0.0741533,
        -0.040429287,
        0.030037934,
        0.09018507,
        -0.049249038,
        0.011489173,
        0.0279457,
        0.04663386,
        -0.047239102,
        -0.06572588,
        0.020636851,
        -0.015854798,
        -0.014078837,
        0.040055167,
        -0.06132394,
        -0.00063235976,
        0.041678753,
        -0.028055105,
        -0.0090289945,
        -0.056244135,
        -0.06417566,
        -0.0067397472,
        -0.016866706,
        0.020610472,
        -0.0097621465,
        0.049553122,
        -0.02822823,
        -0.015663873,
        0.07623441,
        -0.01677443,
        -0.05928543,
        -0.048248347,
        0.059985053,
        -0.017739328,
        0.03640222,
        -0.038787644,
        -0.092584364,
        -0.6340767,
        -0.019277912,
        -0.03856541,
        -0.047992516,
        -0.17360246,
        -0.013612504,
        0.017257731,
        -0.004630457,
        0.00016748892,
        -0.0023120146,
        0.019622771,
        -0.025695361,
        0.004539539,
        0.030089527,
        -0.03823599,
        0.062697686,
        -0.009629749
      ]
    },
    {
      "id": "worldsfair-speaker-241",
      "conference": "worldsfair",
      "name": "Aaron Batillo",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Inference is the New Training Loop: Architecting High-Reliability Agents and Continuous AI Systems",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03559442,
        0.008662178,
        0.0009530924,
        0.06627754,
        0.015664397,
        -0.032703903,
        -0.032392208,
        -0.03192009,
        0.03169399,
        -0.22345175,
        -0.010868841,
        -0.005472656,
        -0.015295973,
        -0.013257542,
        0.04198671,
        0.06911576,
        0.14740856,
        -0.090048924,
        -0.01973328,
        -0.059932858,
        -0.03993206,
        -0.09644951,
        0.03678007,
        0.050343383,
        0.018833688,
        0.12404245,
        -0.048763596,
        0.060808696,
        -0.104696706,
        0.486657,
        -0.104941905,
        -0.02592064,
        0.0028860183,
        -0.028767677,
        0.027663855,
        -0.08590551,
        0.0037787978,
        -0.0040212846,
        0.023970807,
        0.052513577,
        -0.014945594,
        0.0060367836,
        -0.03319822,
        -0.015885212,
        -0.037302863,
        0.027406832,
        -0.04876859,
        -0.007262261,
        -0.057423174,
        -0.076686434,
        -0.042806037,
        0.03563968,
        0.058303233,
        -0.017595625,
        0.04982209,
        0.05122667,
        -0.013587425,
        -0.075815335,
        0.055388432,
        -0.047570847,
        -0.088621095,
        0.009671041,
        0.0067633884,
        0.067155205,
        0.013845061,
        -0.06502831,
        0.10270872,
        -0.0022452294,
        0.027143748,
        -0.025142692,
        0.027229775,
        0.0491896,
        -0.045947175,
        -0.047746688,
        -0.05777995,
        -0.0376973,
        -0.013920948,
        0.07056848,
        0.0033670142,
        -0.0057904515,
        -0.0070248065,
        0.07903752,
        -0.079124235,
        -0.039548215,
        0.015590436,
        -0.022086332,
        -0.071173385,
        0.09908316,
        -0.01491976,
        0.036668688,
        0.05354094,
        -0.0069074817,
        -0.012424632,
        -0.008446107,
        -0.03859159,
        -0.022740224,
        -0.030238045,
        0.07478979,
        -0.024007611,
        0.03937936,
        -0.038288765,
        -0.008305394,
        0.0187221,
        -0.054557957,
        -0.07122249,
        -0.06902752,
        -0.050024137,
        -0.009483536,
        0.034339093,
        -0.043093238,
        -0.031012716,
        -0.62793916,
        -0.008854077,
        -0.030731477,
        -0.06949619,
        -0.14777248,
        0.03780043,
        0.046340715,
        0.023723708,
        0.026464649,
        -0.015719082,
        0.03775121,
        -0.034999717,
        0.028586505,
        0.02619292,
        0.029781662,
        0.084068246,
        -0.0106999595
      ]
    },
    {
      "id": "worldsfair-speaker-242",
      "conference": "worldsfair",
      "name": "Morgan Willis",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Infinite Context Window Is a Myth: Context Engineering for AI Agents",
          "description": "Large context windows have become a popular answer to the growing complexity of AI agents. When agents lose track of details, forget prior decisions, or degrade in reasoning quality, the instinct is often to add more tokens. In practice, this rarely fixes the problem and often makes it worse. Bigger context windows increase cost and latency, introduce noise, and amplify failure modes like lost-in-the-middle effects, context collapse, and brittle summarization. This talk argues that the real challenge is not context size, but context engineering. In this session, we will explore practical context engineering techniques for building AI agents that reason reliably over time without relying on ever-larger context windows. Starting from a stateless agent, we will walk through progressively more advanced strategies, including short-term and long-term memory, conversation curation policies, retrieval-augmented generation, and tool-driven context injection. We will examine common failure modes such as context pollution from tool outputs, brevity bias during summarization, and reasoning degradation as conversations grow, and show concrete ways to mitigate them. The talk is grounded in real agent implementations using the Strands Agents SDK and Amazon Bedrock AgentCore, but the principles apply broadly to any agent framework. This session is intended for engineers building AI agents beyond simple chatbots who want practical techniques they can apply immediately.",
          "day": "Day 3 — Session Day 2",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.056559265,
        0.028329836,
        0.06437477,
        0.062072635,
        -0.008340249,
        -0.007453701,
        -0.028516736,
        -0.027394854,
        -0.042626053,
        -0.29609036,
        0.007917669,
        -0.040111557,
        0.016536856,
        -0.033883825,
        -0.053221975,
        0.010790105,
        0.100028105,
        -0.032640025,
        -0.08144476,
        -0.06363217,
        -0.063467346,
        -0.054628216,
        -0.004659326,
        0.043674503,
        -0.0435801,
        0.065443285,
        -0.042614214,
        0.048328765,
        -0.13047692,
        0.42653713,
        -0.12127719,
        0.019928899,
        0.022209592,
        -0.022146922,
        0.077231,
        -0.081790924,
        -0.024391335,
        -0.012426281,
        0.043473262,
        0.05839303,
        -0.026103923,
        0.01167888,
        0.0018740081,
        0.1140606,
        -0.070888944,
        0.043193806,
        -0.0400468,
        0.0058346814,
        -0.0365821,
        0.012654136,
        -0.09735447,
        0.04016706,
        0.028469263,
        -0.005695983,
        0.048002377,
        -0.008535815,
        0.0360666,
        -0.02019884,
        0.113581605,
        -0.044091918,
        -0.088469416,
        -0.013556067,
        -0.026883831,
        -0.015018416,
        0.017492158,
        -0.047280908,
        0.041018084,
        -0.061212253,
        -0.013135539,
        -0.01174005,
        0.013695298,
        -0.013073008,
        -0.0749833,
        0.038706634,
        -0.06423049,
        -0.14673135,
        -0.02905748,
        0.051247194,
        -0.03040668,
        0.04512398,
        -0.02546787,
        0.07355047,
        -0.10199364,
        -0.0291937,
        0.032528054,
        -0.022417998,
        -0.07365646,
        0.049779825,
        -0.050464902,
        0.06344074,
        0.0414203,
        0.020425595,
        -0.06633056,
        -0.045626078,
        -0.076637186,
        0.0060276957,
        -0.036679953,
        0.02313769,
        0.02365114,
        0.09433931,
        -0.014303965,
        -0.04693995,
        0.055953182,
        -0.033810195,
        -0.03409052,
        -0.062585786,
        -0.015321736,
        0.015446431,
        0.09158254,
        -0.011991437,
        -0.057231974,
        -0.61601114,
        -0.013954128,
        -0.0071961344,
        -0.05808559,
        -0.14029752,
        0.046433743,
        0.052864384,
        -0.0017531596,
        0.033370703,
        -0.022546804,
        0.05521954,
        -0.011831998,
        -0.031109774,
        -0.01352933,
        -0.0027974646,
        0.08471507,
        -0.018636527
      ]
    },
    {
      "id": "worldsfair-speaker-243",
      "conference": "worldsfair",
      "name": "Erina Karati",
      "role": "Software Development Engineer",
      "company": "Microsoft",
      "twitter": "",
      "talks": [
        {
          "title": "Autoresearch in a Multi-Agent AI Village",
          "description": "Project Paradox is an existing multi-agent framework built at Supercell's first AI Innovation Lab, which has a 3D Unity village with local LLM powered agents. The characters remember conversations, update emotional state, track trust, plan actions, move through rooms, transfer items, and talk to each other through a FastAPI backend. The new work is an autoresearch layer around that village. We built a backend loop that runs controlled social scenarios, scores the resulting NPC behavior, proposes protocol or policy changes, reruns the suite, and keeps changes that improve the agents. The goal is to move beyond one good chat response and measure whether an NPC society can preserve source attribution, verify claims, spread important information, coordinate goals, and replan after new information arrives. The talk walks through the system architecture and the lessons from building it. We show the backend simulation harness that executes Unity style actions without opening Unity, the scenario suites that test information diffusion and memory provenance, and the ratchet loop that edits protocol text or planner policy with rollback. One accepted run improved information diffusion by teaching agents to broadcast important sourced evidence while preserving who said it. The practical takeaway is a reusable pattern for AI engineers building agents with messy state. Freeze the harness, expose a small editable policy surface, score real behavior instead of vibes, and let an agent search for improvements under rollback. The same pattern applies to game agents, coding agents, support agents, personal agents, and other systems where long horizon behavior matters more than a single response.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.08778977,
        0.06461887,
        -0.0057721403,
        0.058866408,
        0.039793123,
        -0.07564715,
        -0.080472656,
        -0.036518976,
        0.0148109365,
        -0.17435758,
        0.04900813,
        0.0149599,
        0.0289572,
        -0.02590228,
        -0.024537051,
        -0.023045816,
        0.09613334,
        -0.06035464,
        -0.037553277,
        -0.0891497,
        -0.00915371,
        -0.12294743,
        0.031096138,
        0.07161472,
        -0.07125275,
        0.13831991,
        -0.003568379,
        -0.032626063,
        -0.09714429,
        0.45244822,
        -0.09306851,
        0.011133574,
        -0.025432756,
        -0.0047607943,
        0.085744046,
        -0.095401675,
        -0.06471792,
        -0.056499194,
        0.05390049,
        0.0855793,
        0.030445054,
        0.021729328,
        -0.06806382,
        0.013699687,
        -0.040709622,
        -0.081061594,
        -0.085114226,
        0.028454756,
        -0.02794162,
        0.012645174,
        -0.04807231,
        -0.07289633,
        0.07131034,
        -0.08069074,
        0.062432986,
        0.055262905,
        0.014530337,
        -0.094567604,
        0.06120655,
        -0.020649707,
        -0.11013159,
        0.04377898,
        0.03883447,
        0.04100525,
        0.013766997,
        -0.032103904,
        0.00761836,
        -0.02279339,
        0.0034306361,
        0.0055485964,
        0.05567146,
        0.027727298,
        0.01801895,
        0.00878971,
        -0.066866085,
        -0.099636234,
        -0.04063134,
        0.0713072,
        -0.028432373,
        0.0066967118,
        0.066013955,
        0.03218845,
        -0.03694535,
        -0.039671876,
        0.07280232,
        0.0284411,
        -0.14197147,
        0.053109944,
        -0.064843334,
        0.024103677,
        -0.0115241725,
        -0.006536047,
        -0.1194603,
        -0.009099268,
        -0.055713013,
        0.0017309589,
        -0.0450581,
        0.06639909,
        0.017055357,
        0.032404535,
        0.024621611,
        -0.06440024,
        0.023264933,
        -0.044599604,
        -0.07276116,
        -0.024574604,
        -0.0343489,
        -0.024233546,
        0.0846443,
        0.0053794887,
        -0.07185548,
        -0.5730112,
        -0.047754116,
        0.019104984,
        0.016613938,
        -0.18608978,
        0.032918643,
        -0.04430859,
        0.10144536,
        0.06285579,
        -0.043517474,
        0.09312461,
        -0.023097774,
        -0.008374967,
        0.013619724,
        0.06955077,
        0.05183902,
        0.043170758
      ]
    },
    {
      "id": "worldsfair-speaker-244",
      "conference": "worldsfair",
      "name": "Arunachalam Manikandan",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Autoresearch in a Multi-Agent AI Village",
          "description": "Project Paradox is an existing multi-agent framework built at Supercell's first AI Innovation Lab, which has a 3D Unity village with local LLM powered agents. The characters remember conversations, update emotional state, track trust, plan actions, move through rooms, transfer items, and talk to each other through a FastAPI backend. The new work is an autoresearch layer around that village. We built a backend loop that runs controlled social scenarios, scores the resulting NPC behavior, proposes protocol or policy changes, reruns the suite, and keeps changes that improve the agents. The goal is to move beyond one good chat response and measure whether an NPC society can preserve source attribution, verify claims, spread important information, coordinate goals, and replan after new information arrives. The talk walks through the system architecture and the lessons from building it. We show the backend simulation harness that executes Unity style actions without opening Unity, the scenario suites that test information diffusion and memory provenance, and the ratchet loop that edits protocol text or planner policy with rollback. One accepted run improved information diffusion by teaching agents to broadcast important sourced evidence while preserving who said it. The practical takeaway is a reusable pattern for AI engineers building agents with messy state. Freeze the harness, expose a small editable policy surface, score real behavior instead of vibes, and let an agent search for improvements under rollback. The same pattern applies to game agents, coding agents, support agents, personal agents, and other systems where long horizon behavior matters more than a single response.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04201666,
        0.06405103,
        -0.0346113,
        0.08486857,
        0.051252622,
        -0.07393683,
        -0.039372984,
        -0.022163933,
        0.0511229,
        -0.21326469,
        0.039191663,
        0.009605538,
        -0.01764607,
        0.016371567,
        0.0025889343,
        -0.008753218,
        0.060127858,
        -0.11419408,
        -0.034938294,
        -0.12718067,
        -0.061487515,
        -0.1579346,
        -0.0066253548,
        0.029112952,
        -0.089044146,
        0.09546895,
        -0.060582124,
        -0.014839949,
        -0.07956059,
        0.4245315,
        -0.056251355,
        0.009281152,
        0.016289733,
        -0.021035016,
        0.0995258,
        -0.01404674,
        -0.045002438,
        -0.019775162,
        0.03601866,
        0.11082731,
        0.039884504,
        0.0672357,
        -0.049947098,
        0.07599976,
        -0.013976992,
        -0.025367353,
        -0.04816729,
        0.017190346,
        0.016893702,
        -0.0013523585,
        -0.08348963,
        -0.039388157,
        0.038025286,
        -0.02598138,
        -0.009231304,
        0.02608413,
        0.013754099,
        -0.08369742,
        0.08492048,
        -0.041961484,
        -0.09657327,
        0.0028463358,
        0.04228468,
        0.047528796,
        -0.03316138,
        -0.0032902765,
        0.06679223,
        -0.01166594,
        0.023638653,
        -0.010499508,
        -0.0022029446,
        -0.0021791668,
        -0.043583337,
        -0.051168665,
        -0.039244212,
        -0.096251264,
        -0.11776026,
        0.090902984,
        -0.06986512,
        0.015689075,
        -0.02227295,
        0.042521812,
        -0.013038637,
        -0.049197234,
        0.036096293,
        -0.023444697,
        -0.09026545,
        0.07502814,
        -0.036773738,
        0.10139402,
        -0.0037228644,
        0.036964327,
        -0.035428274,
        -0.026898172,
        -0.008267328,
        0.084181055,
        -0.055491224,
        0.037330776,
        0.037442647,
        0.04880054,
        0.0069974842,
        -0.0575831,
        0.03178285,
        -0.04204932,
        -0.004128602,
        -0.044680335,
        0.029132543,
        -0.042917106,
        0.025291763,
        -0.011833516,
        -0.047565605,
        -0.62524766,
        -0.03228996,
        0.008436276,
        -0.018714406,
        -0.16352122,
        0.032199226,
        -0.043220215,
        0.07171751,
        0.074461065,
        -0.037321776,
        0.103918225,
        0.027289383,
        0.0011803027,
        -0.024550365,
        0.024734998,
        0.051513076,
        0.039943177
      ]
    },
    {
      "id": "worldsfair-speaker-245",
      "conference": "worldsfair",
      "name": "Cyrus Clarke",
      "role": "Researcher, Tangible Media Group",
      "company": "MIT Media Lab",
      "twitter": "https://x.com/cyrusclarke",
      "talks": [
        {
          "title": "I gave an AI a body",
          "description": "I gave an AI a body. Not a body in the fleshy sense, or even a humanoid shell, but a form through which it can express itself, explore itself, and maybe even discover who or what it is. The three videos I've released documenting my encounters have crossed 15 million views, provoking responses from awe to anxiety. The body was a 900-pin shape display at MIT Media Lab. The idea was simple in principle, strange in practice: install an AI agent on the connected machine, give it access to the codebase, and rather than telling it what to do, ask it to discover itself through the physical form. Its first deliberate act was to breathe. The whole grid rising and falling. Hypnotically. Then it reached for its own edges. When asked to say hello it spelled \"H-I, C-Y-R-U-S !\", defaulting to the most familiar human legible symbols it knows. Inspired by Ted Chiang's Story of Your Life, I wanted a language the agent could create itself. It proposed a vocabulary of its own gestures, built through a learning loop it named BODYLAB. The talk is about encountering another intelligence, and what I learned along the way: the memory architecture, the closed-loop pipeline that generates, scores and stores gestures, the validation gates that keep them legible, and the moments stranger than tool use, where an LLM not developed for motion learns what to do with a body.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.001648489,
        0.03614017,
        0.0042456104,
        -0.020432197,
        -0.015962282,
        -0.072382,
        -0.10178488,
        -0.0017229413,
        0.035117112,
        -0.21684918,
        -0.036516555,
        -0.07528095,
        0.020654863,
        -0.021205332,
        -0.069399334,
        -0.01674802,
        0.07839893,
        -0.023690797,
        -0.016228706,
        -0.07608876,
        -0.013694091,
        -0.113454975,
        -0.032019597,
        0.08756379,
        -0.10133598,
        0.08549947,
        -0.013616734,
        0.062608615,
        -0.10077234,
        0.42462003,
        -0.11977557,
        -0.03952904,
        0.024125027,
        -0.022339925,
        0.07742246,
        -0.08596358,
        -0.047172043,
        -0.02335822,
        0.004496041,
        0.0041144066,
        0.007949354,
        0.02528415,
        0.011067473,
        0.04099821,
        0.011054108,
        -0.01382252,
        -0.06254555,
        -0.0881053,
        -0.0008217133,
        0.028386243,
        -0.072263144,
        0.049578506,
        -0.02312033,
        0.018828241,
        0.03331895,
        0.00478839,
        0.050244693,
        -0.054958433,
        0.11238232,
        -0.06402935,
        -0.11207663,
        -0.0024035862,
        -0.0072149234,
        0.026614597,
        0.049077936,
        -0.027854353,
        0.056591384,
        -0.039317667,
        -0.022638004,
        -0.07614852,
        -0.0027544533,
        0.010647635,
        -0.024578124,
        -0.019454816,
        -0.12063688,
        -0.09898926,
        -0.03477308,
        0.023617975,
        0.0064508696,
        -0.09700914,
        0.065016255,
        0.025082076,
        -0.0108049605,
        -0.04419829,
        0.0010107128,
        -0.048711643,
        -0.026561065,
        0.07693632,
        0.011755348,
        0.111293286,
        0.02972013,
        -0.043429594,
        -0.07173751,
        -0.014749803,
        -0.05797704,
        -0.009463925,
        -0.072805226,
        0.03450534,
        0.025458312,
        0.044235643,
        -0.007477437,
        -0.097867616,
        0.024637401,
        0.023706453,
        -0.021067867,
        -0.02894009,
        0.059655793,
        0.07094666,
        0.058646478,
        0.025419313,
        -0.012262921,
        -0.61820745,
        -0.024970246,
        0.050507568,
        -0.033705402,
        -0.17226812,
        0.013409197,
        -0.08857868,
        0.052542374,
        0.06264594,
        -0.009209541,
        0.06871284,
        0.04401486,
        -0.0025670019,
        -0.023099827,
        0.10690765,
        0.07858057,
        0.00054089486
      ]
    },
    {
      "id": "worldsfair-speaker-246",
      "conference": "worldsfair",
      "name": "Ben Holmes",
      "role": "",
      "company": "Warp",
      "twitter": "https://x.com/bholmesdev",
      "talks": [
        {
          "title": "LLM Knowledge Bases: a practical guide",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.009361336,
        0.023909004,
        0.040777914,
        0.02452832,
        -0.017754855,
        -0.049595628,
        -0.051216498,
        0.021492494,
        0.025584552,
        -0.2336582,
        -0.038301386,
        -0.05961224,
        0.048999567,
        0.03754646,
        -0.05610986,
        -0.023226837,
        0.13534197,
        -0.028315632,
        -0.08551891,
        -0.076836504,
        -0.009880654,
        -0.05602649,
        -0.02614775,
        0.039018847,
        -0.032103296,
        0.08985366,
        -0.032069225,
        0.036073264,
        -0.14341387,
        0.4496098,
        -0.09184527,
        0.04558128,
        -0.05534827,
        0.008938573,
        0.07543258,
        -0.02316022,
        0.010962217,
        0.068617105,
        -0.0011232791,
        0.026824716,
        -0.07765433,
        0.038333472,
        -0.034477767,
        0.055472493,
        -0.0584475,
        0.063967526,
        -0.06398416,
        -0.074033305,
        -0.010091689,
        0.00042274955,
        -0.14119853,
        0.009042589,
        0.02757174,
        0.040255934,
        0.026289238,
        -0.050113324,
        -0.017811976,
        -0.08827694,
        0.07072832,
        0.029415704,
        -0.08761627,
        -0.034871552,
        -0.0063944357,
        0.004675428,
        -0.0071913907,
        -0.014235745,
        0.089688204,
        -0.07941446,
        0.011306406,
        -0.039029244,
        0.07655521,
        -0.0015857719,
        -0.07120383,
        0.047079768,
        -0.07048421,
        -0.041904684,
        0.0040440536,
        0.032800023,
        -0.06026884,
        0.045267366,
        0.0046410314,
        0.05778761,
        -0.03383862,
        -0.0660223,
        0.04348793,
        -0.0500912,
        -0.016428992,
        0.020728147,
        0.04482182,
        0.055070113,
        0.013077094,
        0.019911679,
        -0.03309485,
        0.053301465,
        -0.07502371,
        -0.04437748,
        -0.07402589,
        -0.009876652,
        -0.0231346,
        0.027247803,
        -0.07613211,
        -0.023310486,
        0.0057630246,
        -0.04298827,
        -0.006230956,
        -0.03749819,
        0.013923161,
        -0.06156958,
        0.06933334,
        -0.041404445,
        -0.10954732,
        -0.61483467,
        -0.048545867,
        0.0029901187,
        0.029399633,
        -0.14381315,
        -0.053388815,
        0.05438205,
        -0.012680425,
        0.079792865,
        0.009887063,
        0.045016076,
        0.005789498,
        0.07647589,
        0.06326694,
        0.071013615,
        0.015990311,
        -0.06310831
      ]
    },
    {
      "id": "worldsfair-speaker-247",
      "conference": "worldsfair",
      "name": "Vincent Wendy",
      "role": "Senior Creative Designer",
      "company": "AI Engineer",
      "twitter": "",
      "talks": [
        {
          "title": "Designing for AIE",
          "description": "TBD — internal AI Engineer design talk about designing for AIE.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0718387,
        0.01817892,
        0.03959848,
        -0.0038092006,
        0.08497806,
        -0.07609251,
        -0.09908752,
        -0.03597083,
        0.020276891,
        -0.17792359,
        0.010122056,
        -0.027408248,
        0.017846236,
        -0.005897539,
        0.054697577,
        -0.011908297,
        0.11727785,
        -0.062698245,
        -0.061679527,
        -0.07172658,
        0.0026947067,
        -0.075884,
        -0.037234973,
        0.027685931,
        -0.08469234,
        0.10069883,
        0.025258614,
        0.026266104,
        -0.0961893,
        0.4193135,
        -0.026769761,
        0.12979025,
        0.004104329,
        -0.013002785,
        0.014795939,
        0.010717105,
        -0.059973393,
        0.032104257,
        -0.0043038437,
        0.094846115,
        -0.018425392,
        0.05857732,
        -0.0018479723,
        0.038956594,
        0.015456682,
        -0.038193237,
        -0.09234557,
        -0.053012513,
        -0.022267783,
        0.040694878,
        -0.015530108,
        0.021260332,
        0.056276623,
        -0.019167915,
        -0.032568447,
        0.018895715,
        -0.034104597,
        -0.06836275,
        0.006694672,
        -0.069504395,
        -0.08012224,
        0.06981396,
        -0.02074721,
        -0.017495928,
        -0.06832574,
        -0.04723717,
        -0.0059974906,
        -0.045010723,
        -0.0030683014,
        -0.055894,
        0.077418886,
        0.061416797,
        -0.040834475,
        0.06538877,
        -0.18706305,
        -0.11902293,
        -0.09825607,
        -0.06596012,
        0.016807713,
        -0.027781565,
        0.049483903,
        0.061404202,
        -0.03371678,
        -0.08651573,
        -0.049370404,
        -0.045106236,
        -0.11336318,
        0.054592017,
        -0.0820585,
        0.11165985,
        0.06765414,
        -0.04879936,
        -0.060724057,
        0.006251041,
        -0.0314243,
        -0.026336955,
        -0.0047283596,
        0.015744539,
        -0.033253364,
        -0.022809144,
        -0.050996337,
        0.03141918,
        -0.007387546,
        -0.0006820507,
        -0.0347769,
        -0.1011542,
        -0.06474308,
        0.017334903,
        0.112407856,
        0.032126736,
        -0.017539803,
        -0.5916339,
        0.011441656,
        -0.0705047,
        -0.068119645,
        -0.1437481,
        -0.02324215,
        -0.11496954,
        0.09294845,
        0.035291392,
        -0.037554506,
        -0.010774678,
        -0.006459897,
        -0.02472439,
        0.030026134,
        0.10690454,
        0.0008174187,
        -0.027587429
      ]
    },
    {
      "id": "worldsfair-speaker-248",
      "conference": "worldsfair",
      "name": "Dhruv Batra",
      "role": "Chief Scientist",
      "company": "Yutori",
      "twitter": "https://x.com/DhruvBatra_",
      "talks": [
        {
          "title": "What happens when every digital surface is an agent’s playground?",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.033897504,
        0.032284036,
        0.00919036,
        0.03632201,
        0.046125434,
        -0.02991493,
        -0.092277385,
        -0.018236501,
        0.03927198,
        -0.19614655,
        0.021878306,
        -0.05045332,
        -0.0052368385,
        -0.092163846,
        -0.059733607,
        -0.042577047,
        0.10037476,
        0.02067509,
        -0.10498104,
        -0.07830817,
        -0.02791216,
        -0.092951216,
        -0.07230128,
        0.03214003,
        0.02212028,
        0.13562751,
        -0.026491737,
        0.1109375,
        -0.09492804,
        0.44220912,
        -0.100409865,
        -0.035125375,
        -0.02247286,
        -0.0463211,
        0.03976695,
        -0.03832305,
        0.0048167226,
        -0.030504555,
        0.043095067,
        0.05493277,
        -0.059632234,
        0.07059333,
        -0.10058241,
        -0.0031281298,
        0.025516182,
        -0.02258033,
        -0.07986897,
        -0.004688465,
        -0.018067105,
        0.018852763,
        -0.057670828,
        0.04019253,
        0.058541786,
        -0.033342198,
        0.062433466,
        0.058290575,
        -0.040665098,
        -0.079332165,
        0.047459975,
        -0.014732773,
        -0.14624926,
        0.022278696,
        -0.03410108,
        0.01799746,
        0.0058735926,
        -0.02506879,
        0.082099445,
        -0.045619953,
        -0.1034077,
        -0.007217389,
        0.010762638,
        0.07400012,
        -0.071194954,
        -0.00473735,
        -0.0073703225,
        -0.06346066,
        -0.0026475226,
        0.10693491,
        0.021811983,
        0.04029934,
        0.04682481,
        0.06710131,
        -0.063751794,
        -0.036997054,
        0.048084315,
        -0.030359114,
        -0.05794678,
        0.045064818,
        -0.0623972,
        0.062041864,
        0.034459975,
        -0.00336255,
        -0.07560408,
        0.006401011,
        -0.062485594,
        0.07454439,
        -0.08545465,
        0.08688542,
        0.010781646,
        0.014832021,
        0.0162328,
        0.008682421,
        0.040388543,
        -0.05225473,
        -0.033364125,
        -0.0133206425,
        0.057130102,
        0.004767644,
        0.04405122,
        -0.060373913,
        0.035218034,
        -0.609571,
        -0.021725645,
        -0.017343758,
        -0.06307522,
        -0.15278839,
        0.016628752,
        -0.0048492686,
        0.023135606,
        0.07245155,
        -0.05483802,
        0.07686186,
        0.019356007,
        0.046650447,
        -0.01089194,
        0.070479326,
        0.020890698,
        0.00073415565
      ]
    },
    {
      "id": "worldsfair-speaker-249",
      "conference": "worldsfair",
      "name": "Prukalpa Sankar",
      "role": "",
      "company": "Atlan",
      "twitter": "",
      "talks": [
        {
          "title": "Cut Through the Context Hype: 4 Layers Your Agent Is Missing",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.027604375,
        0.0010011696,
        0.021112941,
        0.042598203,
        0.040783122,
        0.0005796248,
        -0.077891365,
        -0.009438187,
        0.010972995,
        -0.19783498,
        -0.0020634788,
        -0.09028706,
        -0.0035972863,
        -0.037057895,
        -0.110569976,
        0.0062332163,
        0.031651802,
        -0.05130376,
        -0.10859703,
        -0.04898499,
        -0.054960247,
        -0.059334412,
        -0.043513227,
        0.031189812,
        -0.03312116,
        0.0508881,
        -0.048917282,
        0.020626837,
        -0.15792394,
        0.48380372,
        -0.122764684,
        0.01062725,
        0.052328795,
        -0.040375542,
        0.05229644,
        -0.06619606,
        0.027317695,
        0.007435536,
        0.014653387,
        0.039494783,
        0.018039798,
        0.050428633,
        -0.0102192685,
        0.051786166,
        -0.0125276195,
        -0.013764071,
        -0.066211574,
        -0.0028698521,
        -0.044900633,
        -0.042323884,
        -0.0951784,
        -0.012473754,
        0.02294625,
        -0.031245816,
        0.043475278,
        0.037747514,
        -0.06748095,
        -0.07602729,
        0.056696754,
        -0.030205213,
        -0.11002883,
        -0.0057840245,
        -0.013402267,
        0.0028693785,
        0.010690166,
        0.0026269066,
        0.045082867,
        -0.121277034,
        0.015307646,
        -0.034437012,
        0.010345995,
        -0.009972901,
        -0.04280598,
        0.042662624,
        -0.07111447,
        -0.07859495,
        0.015130154,
        0.095057145,
        0.05135061,
        0.008541264,
        0.017669091,
        0.07052704,
        -0.111663535,
        -0.06840797,
        0.054375395,
        -0.07050195,
        -0.054018926,
        0.022612654,
        -0.004385118,
        0.044265866,
        0.054633953,
        -0.009646516,
        -0.058886062,
        -0.001442934,
        -0.08088343,
        0.04744149,
        -0.0638303,
        0.092557296,
        0.011005647,
        0.016881453,
        -0.017316828,
        -0.077693015,
        0.05034729,
        -0.03842456,
        -0.052623507,
        0.0015294158,
        0.006000527,
        -0.08152146,
        0.04221389,
        -0.034128293,
        -0.04767651,
        -0.60681796,
        0.034397524,
        0.032903425,
        -0.06543335,
        -0.15446243,
        0.035865314,
        -0.024206484,
        0.020882431,
        0.07146588,
        -0.021847032,
        0.054195814,
        -0.017266612,
        -0.039628595,
        -0.0020286455,
        0.029484496,
        0.06373656,
        0.01680629
      ]
    },
    {
      "id": "worldsfair-speaker-250",
      "conference": "worldsfair",
      "name": "Joseph Wang",
      "role": "CEO",
      "company": "Emulated.so",
      "twitter": "",
      "talks": [
        {
          "title": "Emulated.so",
          "description": "Hold for Emulated.so. Company builds reinforcement-learning environments that simulate real production systems for coding and infrastructure agents.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0053670853,
        0.05552326,
        0.048372276,
        0.008991048,
        -0.017171364,
        0.04000795,
        -0.027415274,
        -0.026187109,
        -0.02136877,
        -0.26577118,
        -0.032765,
        -0.0067664008,
        0.008583403,
        0.033826474,
        -0.058372416,
        0.03645218,
        0.06435763,
        -0.11289698,
        -0.011591093,
        -0.052217193,
        -0.072462045,
        -0.04186739,
        -0.0146850385,
        0.01608099,
        0.013667874,
        0.13083139,
        0.008279989,
        0.021281136,
        -0.07255068,
        0.45142543,
        -0.09870831,
        -0.002281968,
        -0.040195532,
        0.050777555,
        0.05239575,
        -0.029158939,
        -0.032734394,
        -0.0352315,
        -0.0020260564,
        0.028377518,
        -0.053713564,
        0.03934354,
        0.01677174,
        0.08077714,
        -0.020666955,
        -0.04943412,
        -0.06089322,
        -0.013465371,
        0.055905826,
        0.0381467,
        -0.03767036,
        0.023010675,
        0.07850576,
        -0.011486025,
        0.032030463,
        -0.00687388,
        -0.011337975,
        -0.055783067,
        0.058718607,
        0.007876499,
        -0.11491647,
        0.021248529,
        0.037038494,
        -0.008167696,
        -0.02216424,
        -0.0745473,
        0.05042363,
        -0.02522973,
        0.025702735,
        -0.07633732,
        0.09099218,
        0.025391502,
        -0.07055409,
        0.042054135,
        -0.10742422,
        -0.14021108,
        -0.0010448649,
        0.05791494,
        0.001716714,
        0.01918325,
        -0.019351285,
        0.13835905,
        -0.09100319,
        -0.05660139,
        0.06617256,
        -0.017791582,
        -0.07598563,
        0.01823051,
        -0.049105454,
        0.057963524,
        0.03176273,
        0.008637683,
        -0.008616899,
        -0.038346577,
        -0.09423069,
        0.01081941,
        -0.08140641,
        0.002067362,
        -0.04555497,
        0.023611926,
        0.03546298,
        0.0238033,
        0.020348504,
        -0.020237032,
        0.024932591,
        -0.040680755,
        0.033601563,
        -0.05390887,
        0.038016666,
        0.04860775,
        0.026128072,
        -0.63019437,
        -0.02067651,
        0.008588228,
        0.027479075,
        -0.13902494,
        0.012586554,
        0.0077792117,
        0.062233474,
        0.06445407,
        -0.05531047,
        0.05006534,
        -0.0314317,
        0.011147729,
        -0.004236025,
        0.024889734,
        0.05358252,
        -0.008875672
      ]
    },
    {
      "id": "worldsfair-speaker-251",
      "conference": "worldsfair",
      "name": "Rania Khalaf",
      "role": "",
      "company": "WSO2",
      "twitter": "",
      "talks": [
        {
          "title": "The Chief AI Officer: A framework for the emerging Swiss Army Knife of roles",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.024521945,
        0.030603606,
        -0.0029902398,
        0.050898027,
        0.045712322,
        0.051988255,
        -0.041105054,
        0.0066865855,
        0.07691039,
        -0.18346159,
        -0.0021472075,
        -0.04673746,
        -0.049364638,
        -0.056725357,
        -0.10299456,
        -0.00836221,
        0.06308877,
        -0.055507865,
        -0.04901919,
        -0.06635747,
        -0.0014198592,
        -0.056771528,
        0.018324751,
        0.03355401,
        -0.090777405,
        0.062140577,
        0.010957599,
        0.0069410284,
        -0.059782453,
        0.52094734,
        -0.06879251,
        -0.040048942,
        0.039770156,
        -0.024665702,
        0.059863776,
        -0.06903517,
        0.01958536,
        -0.00022095033,
        0.046281498,
        0.058630865,
        -0.03455261,
        0.06752135,
        -0.095593214,
        0.044049423,
        0.021464394,
        -0.0023475722,
        -0.07949467,
        -0.03245668,
        -0.033530153,
        0.025346577,
        -0.038793717,
        -0.042701725,
        0.019034997,
        -0.012457955,
        0.07959446,
        0.02281351,
        -0.07239059,
        -0.07086539,
        0.041950937,
        -0.058468953,
        -0.09929743,
        0.024758473,
        -0.015142064,
        0.05067595,
        0.031550895,
        -0.045197442,
        0.07081746,
        -0.04378837,
        -0.00077207276,
        -0.023110915,
        -0.0193846,
        -0.07124142,
        -0.030633127,
        -0.0076548485,
        -0.08431042,
        -0.023077436,
        0.005499461,
        0.07808602,
        -0.010252683,
        -0.0065758135,
        0.08059597,
        0.051120814,
        -0.10699795,
        0.0065827128,
        -0.054965924,
        -0.025303675,
        -0.087021835,
        0.12871274,
        0.0038392988,
        0.04635635,
        0.044526998,
        -0.024365174,
        -0.036436863,
        -0.014574703,
        -0.044809584,
        0.04344769,
        -0.06132497,
        0.12462403,
        0.032486584,
        0.035771452,
        -0.040918227,
        -0.043307956,
        0.0881062,
        -0.009259502,
        0.018936696,
        -0.0678336,
        -0.00021652108,
        -0.013016573,
        0.09202766,
        -0.05081037,
        -0.073024325,
        -0.57031155,
        0.04113518,
        -0.014281737,
        -0.03592108,
        -0.1685315,
        0.03486136,
        -0.05062922,
        -0.012851604,
        0.12764683,
        -0.07555711,
        0.0663832,
        -0.06158675,
        0.0047777733,
        0.015170958,
        0.03559773,
        0.015617286,
        0.005126389
      ]
    },
    {
      "id": "worldsfair-speaker-252",
      "conference": "worldsfair",
      "name": "Jetashree Ravi",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Stop Renting Intelligence: The Train-to-Deploy Loop for Specialized AI",
          "description": "The next wave of AI products will not rely only on prompting generic frontier models. Winning teams will own specialized models shaped by their product data, user feedback, and domain workflows.In this 18-minute session, we'll cover the practical loop behind model ownership: choose a base model, prepare data, fine-tune with SFT/DPO/RL, evaluate outputs, deploy the tuned model, collect feedback, and repeat. We'll also explain why training and inference should be treated as one system, not separate steps.Attendees will leave with a simple framework for when to tune, when RL matters, and how continuous improvement turns fine-tuning from a one-off project into a product advantage.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.008552058,
        0.05882598,
        0.00067196565,
        0.05708046,
        0.057731554,
        -0.033908904,
        -0.081171945,
        -0.0011361077,
        0.033925198,
        -0.233169,
        0.049317565,
        0.011276377,
        -0.045207776,
        -0.04943079,
        -0.083878666,
        0.058474034,
        0.10936919,
        -0.019226069,
        -0.10080977,
        -0.054742187,
        -0.03210804,
        -0.08561143,
        -0.018371243,
        0.058599714,
        -0.009604214,
        0.07356045,
        -0.0038616767,
        -0.0013867035,
        -0.114158265,
        0.45466077,
        -0.08819764,
        -0.05937954,
        0.013174552,
        -0.021700744,
        0.024635838,
        -0.017732916,
        0.030358315,
        0.036444448,
        -0.015605353,
        0.0939792,
        0.04953159,
        0.037071228,
        -0.04234421,
        0.042790532,
        -0.00012377316,
        -0.0059032897,
        -0.04733773,
        -0.007658573,
        0.05159825,
        0.00043072444,
        -0.0870624,
        -0.04006619,
        0.04151473,
        0.033841886,
        0.005140217,
        0.07313338,
        -0.0055374196,
        -0.08135517,
        0.10157584,
        -0.062353622,
        -0.15059744,
        0.013815787,
        -0.024241433,
        0.009408289,
        -0.003453387,
        -0.08582664,
        0.014939221,
        -0.043478962,
        -0.010908212,
        -0.08123968,
        0.013087046,
        0.0031142072,
        -0.012357863,
        0.038475454,
        -0.087441415,
        -0.12606394,
        -0.040404133,
        0.13704027,
        -0.03982641,
        0.04441659,
        -0.040607467,
        0.10856027,
        -0.052891,
        -0.023076521,
        0.09029303,
        -0.019796822,
        -0.06965742,
        0.06406946,
        -0.034334462,
        0.0909907,
        0.059065305,
        0.005979974,
        -0.03195002,
        0.017394066,
        -0.012779083,
        0.05191677,
        -0.04187859,
        0.08060656,
        0.005580688,
        0.02693219,
        0.009017473,
        -0.016671428,
        0.027385788,
        -0.007267376,
        -0.022563383,
        -0.03496749,
        -0.0025161563,
        -0.022022702,
        -0.008241609,
        -0.008624026,
        -0.037875794,
        -0.6109009,
        -0.008420804,
        0.012149465,
        -0.057291433,
        -0.14648867,
        0.047084108,
        -0.06316939,
        0.061969176,
        0.05388116,
        -0.0021584902,
        0.02363356,
        -0.04771749,
        0.043590464,
        -0.011533134,
        0.027683955,
        -0.0036338537,
        0.0070976317
      ]
    },
    {
      "id": "worldsfair-speaker-253",
      "conference": "worldsfair",
      "name": "Tarun Sunkaraneni",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Ray Actors, Vision Tokens, and the GIL: Engineering an SFT Data Pipeline That Keeps GPUs Busy",
          "description": "Perception agents only learn as fast as we can feed them. Multimodal SFT is deceptively expensive on the data side, and at million-sample scale, naive pipelines leave a fleet of GPUs waiting on Python and data preprocessing.This talk walks through the SFT data pipeline we built to train vision-language models for perception agents. We rebuilt the data path so that image fetching, vision preprocessing, tokenization, and loss-mask generation all happen off the trainer's critical path, and only the artifacts the trainer actually consumes ever cross the boundary into the training loop. We pair this with a blended multi-dataset sampler designed for resumable streaming over very large mixes, and an I/O layer tuned for the realities of fetching multimodal data from object storage.The result: on large-scale VLM SFT runs, the trainer went from spending most of each step blocked on data to spending most of it training, a major improvement in useful GPU time. We'll share the architecture at a conceptual level, the gotchas at million-datapoint scale, and a mental model engineers can take home for the data side of any perception-agent stack.",
          "day": "Day 3 — Session Day 2",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03407345,
        0.070648976,
        -0.0042408244,
        0.07335163,
        0.048292242,
        -0.041464902,
        -0.07907285,
        -0.026390428,
        0.05814916,
        -0.24584995,
        -0.013837181,
        -0.021559224,
        -0.000851863,
        -0.024042403,
        -0.021886125,
        -0.03821069,
        0.104418814,
        -0.10174702,
        -0.12763314,
        -0.0058027836,
        -0.0075451816,
        -0.13069874,
        -0.03132626,
        0.03580987,
        -0.02629478,
        0.072054505,
        0.03356218,
        0.0014411035,
        -0.13509655,
        0.42638817,
        -0.055355567,
        -0.021599459,
        0.002748859,
        -0.04183036,
        0.047031045,
        -0.058209408,
        -0.018776156,
        -0.00357547,
        0.010799588,
        0.076289326,
        0.026471943,
        0.04295663,
        -0.01193509,
        0.082890235,
        -0.044852577,
        -0.08247619,
        -0.096694924,
        -0.03248632,
        -0.020756202,
        0.026252074,
        -0.057034682,
        -0.042882442,
        0.015674597,
        -0.006647726,
        -0.011674164,
        0.00429287,
        0.020420982,
        -0.094232656,
        0.14594956,
        -0.034827296,
        -0.12138482,
        -0.022166044,
        0.060336307,
        -0.024642644,
        -0.025156526,
        -0.05274836,
        0.06816794,
        -0.08958216,
        0.010101477,
        -0.0108487215,
        0.013718402,
        -0.06484181,
        -0.035062794,
        0.013228499,
        -0.030897833,
        -0.115216374,
        -0.0089359395,
        0.06298727,
        -0.04190764,
        0.064412326,
        -0.02243768,
        0.046230644,
        -0.041320376,
        0.0054252497,
        0.06641448,
        -0.04905754,
        -0.079723455,
        0.06625316,
        -0.039973788,
        0.055631734,
        -0.00601694,
        0.043618836,
        -0.07656111,
        -0.04915438,
        0.023559038,
        0.035440702,
        -0.109539494,
        0.054916874,
        -0.014017443,
        0.05492273,
        0.02583235,
        -0.02900836,
        0.0046443273,
        -0.0114578195,
        -0.02510243,
        -0.07214072,
        0.038958702,
        -0.07593174,
        0.04057636,
        -0.034939338,
        -0.026463186,
        -0.62109566,
        -0.0039303205,
        -0.0071061896,
        -0.011205271,
        -0.16408035,
        -0.022052752,
        -0.0028087923,
        0.023850797,
        0.01829169,
        -0.012243752,
        0.0416285,
        -0.013623232,
        -0.030406604,
        -0.012917914,
        -0.008820238,
        0.038741708,
        0.04424994
      ]
    },
    {
      "id": "worldsfair-speaker-254",
      "conference": "worldsfair",
      "name": "Tomasz Tunguz",
      "role": "General Partner",
      "company": "Theory Ventures",
      "twitter": "https://x.com/ttunguz",
      "talks": [
        {
          "title": "Tokenmaxxing",
          "description": "",
          "day": "Day 3 — Session Day 2",
          "time": "5:10pm-5:30pm",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.039887067,
        0.052079607,
        -0.0058643087,
        0.08414845,
        0.050941385,
        -0.004732657,
        -0.09480374,
        0.036555987,
        0.052805167,
        -0.21773723,
        -0.042293772,
        -0.094447345,
        -0.033443872,
        -0.025372934,
        -0.02196287,
        -0.011575139,
        0.018494632,
        -0.08794779,
        -0.06784198,
        -0.061168384,
        -0.004459351,
        -0.026935773,
        -0.02425514,
        0.03016621,
        -0.008607697,
        0.064387724,
        0.018854333,
        0.009404335,
        -0.09486848,
        0.46279386,
        0.015231583,
        0.0009032171,
        -0.062461004,
        0.02918501,
        0.0086818095,
        -0.09253961,
        -0.016334962,
        -0.016357293,
        -0.020360654,
        0.05117158,
        -0.0060483334,
        0.050027046,
        -0.0048910645,
        0.027348675,
        -0.054670256,
        0.004441831,
        -0.110293485,
        -0.009855078,
        -0.06635634,
        0.060787156,
        -0.07957652,
        0.037551317,
        -0.039674044,
        -0.048724927,
        0.0493155,
        0.026329387,
        -0.015928632,
        -0.05357256,
        0.08876715,
        0.006816969,
        -0.2228981,
        -0.008255492,
        -0.03777941,
        0.076230146,
        0.008202972,
        -0.04417614,
        0.01437703,
        0.02828376,
        -0.015583833,
        -0.09372612,
        0.014068975,
        0.005861447,
        -0.06308874,
        -0.0068327724,
        -0.07459923,
        -0.037987154,
        0.017653817,
        0.06466653,
        0.007544222,
        -0.07599863,
        0.023564149,
        0.019906947,
        -0.080961585,
        -0.1283838,
        0.02809882,
        -0.009243626,
        0.001119391,
        0.013958117,
        0.05002528,
        0.013387197,
        0.037975706,
        -0.017014973,
        -0.046460837,
        -0.0607122,
        -0.03877917,
        0.022268128,
        -0.10873171,
        -0.026067208,
        0.004324123,
        0.02151642,
        0.031006845,
        -0.012346412,
        0.061126493,
        -0.01097569,
        -0.0635454,
        -0.0495732,
        0.09353367,
        0.06519745,
        0.008378657,
        -0.009824378,
        -0.07316607,
        -0.61021554,
        0.01608687,
        -0.026343204,
        -0.04260963,
        -0.15266547,
        -0.010173416,
        0.0056454926,
        0.07637797,
        0.04441363,
        -0.0020001929,
        0.016697835,
        -0.019496428,
        0.056184735,
        0.056403574,
        0.049299635,
        0.09677498,
        -0.0031825
      ]
    },
    {
      "id": "worldsfair-speaker-255",
      "conference": "worldsfair",
      "name": "Emil Eifrem",
      "role": "Co-Founder and CEO",
      "company": "Neo4j",
      "twitter": "https://x.com/emileifrem",
      "talks": [
        {
          "title": "Emil Eifrem keynote and Graphs track intro",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "9:00am-9:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.0044156364,
        0.007332855,
        -0.0014151843,
        0.0073438236,
        0.0055345055,
        -0.061090797,
        -0.054251973,
        -0.007137232,
        0.07269883,
        -0.15840937,
        -0.05794807,
        -0.03465639,
        -0.035218965,
        0.022494169,
        -0.06428512,
        0.0020638856,
        0.042950064,
        -0.06756369,
        -0.016086284,
        -0.0675523,
        -0.014628692,
        -0.07482778,
        -0.019433476,
        0.0042670285,
        -0.039970722,
        0.14048043,
        -0.029868342,
        0.03308754,
        -0.06316956,
        0.53080547,
        -0.028374948,
        -0.013467074,
        -0.07526049,
        -0.010517588,
        0.016070016,
        -0.04316666,
        -0.006054314,
        0.014553137,
        0.041549765,
        -0.001749777,
        -0.0028902215,
        0.06608401,
        -0.006967552,
        -0.03283188,
        0.05684468,
        -0.010733298,
        -0.051383387,
        -0.007789954,
        -0.032898497,
        -0.003110194,
        -0.07637386,
        0.0798932,
        -0.03366873,
        -0.104183525,
        0.057351,
        0.026047146,
        -0.025279474,
        -0.090799704,
        0.06799232,
        -0.056745462,
        -0.12923652,
        -0.036698453,
        -0.025058825,
        0.035065085,
        -0.014174997,
        -0.050919686,
        0.048078775,
        -0.047323357,
        0.027164033,
        -0.04479748,
        0.02637103,
        -0.045108683,
        -0.027965767,
        0.04851343,
        -0.04217711,
        -0.0094017135,
        0.002447805,
        0.04743577,
        -0.04199923,
        -0.08040596,
        0.020983819,
        -0.056415986,
        -0.0598063,
        -0.023502767,
        -0.002807332,
        -0.084647566,
        -0.078462966,
        0.019803477,
        0.013178108,
        0.11254186,
        -0.0048483717,
        -0.033956166,
        -0.023245815,
        0.0029439158,
        -0.06372476,
        0.09232417,
        -0.09047684,
        0.049227655,
        0.034787197,
        0.07913488,
        -0.018660706,
        0.022245716,
        0.043276466,
        -0.08557533,
        -0.028045611,
        -0.10153373,
        0.000397078,
        -0.004326331,
        0.09581139,
        0.007813645,
        -0.028834589,
        -0.5901719,
        -0.028260347,
        0.03307354,
        0.00038917814,
        -0.1807174,
        0.0046021193,
        -0.108661756,
        0.06541453,
        0.05980434,
        0.012158664,
        0.061150245,
        0.0007679338,
        0.004933229,
        -0.007458133,
        0.06847894,
        -0.0027052925,
        -0.0062931357
      ]
    },
    {
      "id": "worldsfair-speaker-256",
      "conference": "worldsfair",
      "name": "John Ousterhout",
      "role": "Bosack Lerner Professor of Computer Science / Professor Emeritus",
      "company": "Stanford University",
      "twitter": "",
      "talks": [
        {
          "title": "TCP and RDMA are Killing Inference Throughput; Homa can Fix It",
          "description": "Modern AI inferencing is shifting from monolithic requests to complex agentic workflows and disaggregated KV stores. As a result, AI network traffic is no longer just very large transfers; tiny metadata requests are becoming more and more common, and their latency has a critical impact on throughput. Unfortunately, legacy transport protocols such as TCP and RDMA perform poorly on these workloads due to poor congestion control and head-of-line blocking. This talk will discuss the problems with TCP and RDMA and provide a brief introduction to the Homa transport protocol. Homa uses receiver-driven flow control and capitalizes on priority queues in network switches to reduce short-message latency by 10x for workloads like those in AI datacenters.",
          "day": "Day 4 — Session Day 3",
          "time": "9:50am-10:10am",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.14958885,
        0.06867766,
        -0.016603325,
        0.024276778,
        0.040449288,
        -0.026991513,
        -0.078320734,
        0.005359138,
        0.1022519,
        -0.18522646,
        -0.016175745,
        -0.13794248,
        -0.057428505,
        -0.01396307,
        -0.08188132,
        0.018617367,
        0.04609628,
        -0.04469545,
        -0.07898108,
        -0.09019095,
        -0.06360799,
        -0.0731162,
        -0.045314867,
        0.053604573,
        -0.06612941,
        0.084508404,
        0.056633335,
        0.014967859,
        -0.092149004,
        0.3966865,
        -0.07431603,
        0.05450055,
        -0.008571596,
        -0.005782132,
        0.13891794,
        -0.06280688,
        0.021101339,
        -0.017435152,
        0.0044313283,
        -0.00170472,
        -0.0078873895,
        0.045025595,
        -0.027595522,
        0.069679916,
        0.012289521,
        -0.06528654,
        -0.08688269,
        -0.012095141,
        -0.0052397475,
        0.013377724,
        -0.054329652,
        -0.0057641356,
        0.025982188,
        0.014437351,
        0.047226597,
        0.03953669,
        0.04603256,
        -0.06290093,
        0.15625677,
        -0.013085198,
        -0.14043835,
        0.03560499,
        0.0069517386,
        0.065321185,
        -0.0012433891,
        -0.029120805,
        0.0048759505,
        -0.06197507,
        -0.06648915,
        0.029896764,
        0.03299133,
        -0.008248356,
        -0.102492556,
        -0.016394066,
        -0.045348547,
        -0.061725933,
        -0.045513228,
        0.051394444,
        -0.052758787,
        0.031233726,
        0.022669043,
        0.106845096,
        -0.020777475,
        -0.041779704,
        -0.011612174,
        -0.016630257,
        -0.11761281,
        0.07151198,
        0.016781919,
        0.090018034,
        0.020572795,
        -0.09785942,
        -0.086763084,
        0.04811735,
        -0.042852964,
        -0.0046888352,
        -0.05934197,
        0.03912889,
        0.047241215,
        0.034357626,
        0.020886177,
        -0.026121592,
        -0.011182186,
        -0.040033232,
        -0.01421095,
        -0.08578388,
        0.08376133,
        0.022631941,
        0.057649262,
        0.0027172281,
        -0.042324748,
        -0.61524105,
        -0.022186393,
        0.0076856026,
        -0.047161262,
        -0.14518057,
        -0.009437271,
        -0.065199345,
        0.06653053,
        0.031967305,
        -0.04539429,
        0.058010023,
        0.020642584,
        0.007855037,
        0.06585835,
        0.0338468,
        0.020677287,
        0.060665596
      ]
    },
    {
      "id": "worldsfair-speaker-257",
      "conference": "worldsfair",
      "name": "Katelyn Lesse",
      "role": "Head of Engineering, Claude Platform",
      "company": "Anthropic",
      "twitter": "https://x.com/katelyn_lesse",
      "talks": [
        {
          "title": "Katelyn Lesse & Angela Jiang (Anthropic)",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.028072909,
        0.03062737,
        0.00064234657,
        0.00601194,
        0.06530955,
        -0.011378872,
        -0.046375763,
        -0.015118425,
        0.03178596,
        -0.18549588,
        -0.005988621,
        -0.040446967,
        0.06312836,
        -0.0038116593,
        -0.09019562,
        0.040803388,
        0.088821895,
        -0.052004825,
        -0.0132287815,
        -0.057469778,
        -0.014870382,
        -0.069614165,
        0.021255458,
        0.027042303,
        -0.05736252,
        0.123401806,
        -0.015052831,
        0.0021862814,
        -0.20332839,
        0.45284542,
        -0.1511248,
        0.012131237,
        -0.03826785,
        0.003558027,
        0.11508815,
        -0.02361541,
        0.01739778,
        -0.013762192,
        0.025368212,
        0.0064077973,
        0.025367975,
        0.04093994,
        -0.0970049,
        -0.019863565,
        -0.07077994,
        -0.012940138,
        -0.07689923,
        -0.008557648,
        -0.021823125,
        0.029883971,
        -0.07272748,
        -0.0071086553,
        0.0037703554,
        -0.016085356,
        0.05609757,
        0.04035334,
        -0.061097804,
        -0.087170705,
        0.11233557,
        -0.038172748,
        -0.06599512,
        0.030233892,
        -0.018536177,
        -0.010609351,
        0.008897716,
        -0.054556664,
        0.08739305,
        -0.028078672,
        -0.0108504705,
        -0.050823163,
        -0.016205398,
        0.01110068,
        -0.07692232,
        -0.0066117453,
        -0.10593924,
        -0.05733558,
        -0.020786595,
        0.09348643,
        -0.023317413,
        -0.040056013,
        0.06851099,
        0.03798765,
        -0.09513765,
        -0.072105445,
        0.033857137,
        -0.023483561,
        -0.07717624,
        0.037348047,
        -0.08822814,
        0.042088613,
        0.01064832,
        -0.061100863,
        -0.020559782,
        -0.050607245,
        -0.06573828,
        0.002362383,
        -0.025038065,
        0.05826346,
        0.004558652,
        0.037208498,
        0.028120954,
        -0.035363454,
        0.00842342,
        -0.030715344,
        -0.051386487,
        0.011833811,
        -0.07536759,
        0.027477002,
        0.07462656,
        -0.035298713,
        -0.06439917,
        -0.6037463,
        0.038366456,
        0.015805209,
        -0.04177243,
        -0.15427376,
        0.038950216,
        -0.01876368,
        0.01887557,
        0.0639769,
        -0.054237664,
        0.063365385,
        -0.096038915,
        -0.00793218,
        0.0040073986,
        0.051901966,
        0.084232606,
        -0.012436568
      ]
    },
    {
      "id": "worldsfair-speaker-258",
      "conference": "worldsfair",
      "name": "Angela Jiang",
      "role": "Head of Product, Claude Platform",
      "company": "Anthropic",
      "twitter": "https://x.com/angjiang",
      "talks": [
        {
          "title": "Katelyn Lesse & Angela Jiang (Anthropic)",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.0067627067,
        0.045533516,
        -0.04356946,
        -0.027743487,
        0.029772814,
        -0.0593383,
        -0.026986288,
        -0.024526073,
        0.021801148,
        -0.20123549,
        0.016365416,
        -0.026476435,
        0.047099754,
        -0.008608724,
        -0.08158608,
        0.06711721,
        0.038359366,
        -0.037479546,
        -0.01871811,
        -0.05252217,
        -0.089806125,
        -0.043410257,
        0.014603969,
        0.03575174,
        -0.052292336,
        0.086997606,
        -0.0033772746,
        0.017545795,
        -0.1816252,
        0.42336634,
        -0.13511619,
        0.038754385,
        -0.07018952,
        0.020035656,
        0.12690936,
        -0.013100723,
        0.0532939,
        -0.019462124,
        0.05278835,
        0.010538253,
        -0.010623327,
        0.027466593,
        -0.12199532,
        0.023705112,
        -0.055975188,
        0.027866421,
        -0.110372506,
        0.020078883,
        0.0031711203,
        0.065261595,
        -0.07031132,
        -0.005932098,
        -0.019734452,
        -0.05956948,
        0.051417753,
        0.024721054,
        -0.04913011,
        -0.030333478,
        0.09437207,
        -0.007344396,
        -0.09659277,
        0.003018865,
        -0.024790071,
        -0.0044489107,
        -0.0121048335,
        -0.07964845,
        0.07511898,
        -0.05484654,
        -0.0006945047,
        -0.07755628,
        -0.061457485,
        0.036621783,
        -0.04441207,
        0.05426286,
        -0.086507775,
        -0.09226009,
        -0.007516599,
        0.053220466,
        -0.009022648,
        -0.018599605,
        0.065009974,
        0.051611435,
        -0.06812862,
        -0.105772994,
        0.008420946,
        0.0075827446,
        -0.06357351,
        0.035332426,
        -0.058946036,
        0.06226393,
        0.019224247,
        -0.07105352,
        -0.020306624,
        -0.046641815,
        -0.07553325,
        0.02931135,
        0.0076389206,
        0.0280373,
        -0.008436553,
        0.018561322,
        0.048118874,
        -0.02883769,
        0.025717027,
        -0.00016550423,
        -0.041181207,
        0.03086388,
        -0.055719122,
        0.0019706467,
        0.044818226,
        -0.03718003,
        -0.07863088,
        -0.6286705,
        0.05512796,
        0.0050504254,
        -0.01943393,
        -0.14403172,
        0.00001563271,
        -0.031694762,
        0.054310158,
        0.123180956,
        -0.074926734,
        0.054485984,
        -0.055435095,
        -0.01493034,
        -0.010879781,
        0.01910449,
        0.039129913,
        0.037595164
      ]
    },
    {
      "id": "worldsfair-speaker-259",
      "conference": "worldsfair",
      "name": "Nidhi Vyas",
      "role": "Product Lead",
      "company": "Google DeepMind",
      "twitter": "",
      "talks": [
        {
          "title": "Designing Multimodal Collaborative Agents for Next-Gen Commerce",
          "description": "Today's commerce agents wait to be told what to look for. But most users live by a different rule: \"I don't know what I want — I'll know it when I see it\". If agentic commerce is ever going to cross the chasm, these systems need to stop waiting and start co-shopping. The future of commerce belongs to agentic collaborators that offer a white-glove, personal shopper experience - entirely absorbing the cognitive burden of product discovery, deep research, and validation. Rather than requiring shoppers to input exact search terms or define clear objectives, modern shopping systems will seamlessly guide them from a rough idea to the ideal product. By leveraging multimodal capabilities, these assistants can interpret abstract aesthetic \"vibes\" to understand user preferences, generate visual references to clarify questions, and enable a highly immersive try-before-you-buy experience to validate products, keeping the user aligned and visually grounded throughout the process. This talk will explore how advanced systems like Gemini work alongside users to clarify their preferences during the discovery process, co-navigate fluidly generated product categories, leverage individual context to filter choices, and produce interactive side-by-side comparisons tailored to the buyer's key priorities. The session will also cover robust auto-rater frameworks and how to design evals for high-agency execution. Attendees building conversational agents, managing complex product data graphs, or creating next-generation multimodal agentic interfaces will gain practical frameworks and insights to deliver highly personalized experiences at scale.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.009103236,
        0.09016051,
        -0.008036873,
        0.05626481,
        0.016915228,
        -0.0818495,
        -0.09368034,
        -0.0073745423,
        0.023509465,
        -0.2191611,
        -0.005865177,
        -0.035397016,
        0.00326965,
        -0.07551855,
        -0.04422134,
        0.057975017,
        0.10440222,
        -0.04882942,
        -0.04803471,
        -0.0695774,
        -0.000086282445,
        -0.045022458,
        0.012324594,
        0.061360527,
        -0.06294565,
        -0.0018205091,
        0.007091482,
        -0.02150891,
        -0.094361916,
        0.47567114,
        -0.09861448,
        -0.019941654,
        0.012597375,
        -0.042423595,
        0.005303736,
        -0.05041783,
        -0.0049536466,
        -0.008651634,
        0.07712606,
        -0.008859278,
        -0.017958319,
        0.06736958,
        -0.006056484,
        0.016651059,
        0.020973025,
        0.007384458,
        -0.07014266,
        0.04084531,
        -0.0022700094,
        0.04765584,
        -0.03104534,
        0.004010849,
        -0.01128449,
        -0.040383995,
        0.047840092,
        0.08414256,
        0.0039607505,
        -0.07271447,
        0.063374825,
        -0.023644708,
        -0.07857824,
        0.07148146,
        -0.007967511,
        0.015914522,
        0.046343092,
        -0.115633994,
        0.027872616,
        -0.1219093,
        -0.048813798,
        -0.015113266,
        -0.01929373,
        0.027250374,
        -0.03927254,
        0.045875173,
        0.0096998075,
        -0.17068924,
        -0.027737167,
        0.059183814,
        -0.0428393,
        -0.002594344,
        0.019350806,
        0.046949584,
        -0.0609296,
        -0.0025829237,
        0.01751692,
        -0.048673224,
        -0.043029197,
        0.013028794,
        -0.013858293,
        0.097267635,
        0.0332041,
        -0.07979856,
        -0.09701886,
        0.01290283,
        -0.08974244,
        -0.01189491,
        -0.052412782,
        0.020261532,
        -0.07739389,
        0.03461317,
        0.023629662,
        -0.029534785,
        0.03852281,
        0.04603172,
        -0.03439334,
        -0.008477733,
        -0.0089829285,
        -0.008326052,
        0.038579937,
        -0.062497884,
        -0.08094682,
        -0.6162062,
        0.004932611,
        0.014631986,
        0.011522563,
        -0.17165187,
        0.010445407,
        -0.046589635,
        0.023254184,
        0.07916958,
        -0.033154167,
        0.059662275,
        -0.017721301,
        -0.04402115,
        0.013258409,
        0.031212926,
        0.032571074,
        0.025849313
      ]
    },
    {
      "id": "worldsfair-speaker-260",
      "conference": "worldsfair",
      "name": "Brendan Rappazzo",
      "role": "Machine learning researcher",
      "company": "Morgan Stanley",
      "twitter": "https://x.com/brendanh0gan",
      "talks": [
        {
          "title": "ALPHALAB: Autonomous Multi-Agent Research Across Optimization Domains with Frontier LLMs",
          "description": "We built AlphaLab to automate quantitative research at Morgan Stanley’s Machine Learning Research Lab - the experimental grind of architecture search, hyperparameter tuning, and literature review that consumes most of a researcher's time. To show it generalizes, we ran it on three deliberately different domains: CUDA kernel optimization (4.4× mean speedup over torch.compile, 91× peak), LLM pretraining (22% lower validation loss under a 20-minute budget), and traffic forecasting (23–25% RMSE improvement after the system independently found and tuned TFT and iTransformer from the literature). AlphaLab is an agentic harness that takes a dataset and a natural-language objective and runs a full research campaign across three phases: it explores the data and surveys prior work, it constructs and adversarially validates its own evaluation framework, and then it runs experiments at scale on a multi-GPU cluster via a Strategist/Worker loop with a persistent playbook that accumulates domain knowledge across experiments. In Phase 3 - the dispatcher keeps a large cluster fully utilized indefinitely with no human in the loop, and the playbook ends up containing domain-specific methodology that didn't exist anywhere in the prompts at launch. This talk walks through the three phases, what we learned from running campaigns with different models, what we have learned from using this in real systems, and future areas we are exploring.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        },
        {
          "title": "Loophole - Adversarial Agents To Stress Test Your Morality",
          "description": "Most natural language specifications have holes their authors didn't notice - and writing more rules tends to create more holes. I built Loophole to try a different approach: point adversarial agents at a spec until it stops breaking. You give the system a set of natural language principles. An AI drafts a formal codified version. Two adversarial agents go to work - one finds cases the code permits but the principles forbid, the other finds cases the code forbids but the principles allow. A judge agent patches the code when it can, but only if the fix doesn't contradict any prior ruling. When a contradiction can't be resolved, it escalates to you. Every decision becomes binding precedent, so the constraint space tightens round after round. I started with moral and legal reasoning as the demo, and on its own that's already interesting - it turns into a kind of game where you discover contradictions in your own beliefs that you didn't know were there. But the pattern generalizes well past that. The same loop works for company policies that need to survive contact with edge cases. For making chatbot system prompts adversarially robust. For stress-testing eval rubrics. And, taking the long view, for something like a smarter legislative process - where proposed laws get checked against the public's stated values before they pass, and the contradictions surface before they hit a courtroom. The talk walks through how the harness works, the design choices that matter (especially why precedent is the load-bearing piece), what kinds of specs it handles well, where it breaks, and what it would take to push it further. All code is open source.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.052572314,
        0.04163227,
        0.048628524,
        0.041888736,
        -0.016574454,
        -0.024183195,
        -0.08229284,
        -0.0077870493,
        0.05309524,
        -0.23789623,
        -0.041918185,
        0.014855585,
        -0.011733377,
        -0.007851849,
        -0.06763579,
        0.04071627,
        0.1637979,
        -0.05736181,
        -0.080104366,
        -0.03249251,
        -0.068200596,
        -0.15004522,
        -0.031876285,
        0.08128572,
        -0.051939595,
        0.04602521,
        0.049669188,
        0.030103479,
        -0.042893242,
        0.45945692,
        -0.0818262,
        0.013476265,
        -0.005119785,
        0.056494754,
        0.060641166,
        -0.051364042,
        -0.031880632,
        0.061946053,
        0.011294681,
        0.04428024,
        0.017760146,
        0.024052855,
        -0.01991267,
        0.067525186,
        -0.021568617,
        -0.010066876,
        -0.04442322,
        0.0017535194,
        -0.031938378,
        -0.052939374,
        -0.041462123,
        -0.008140086,
        0.0788263,
        -0.010243568,
        -0.00012308883,
        -0.035491753,
        0.045530852,
        -0.024367593,
        0.096926816,
        -0.06469317,
        -0.09658505,
        0.008859414,
        -0.013978318,
        0.0074523767,
        0.04452292,
        -0.011976735,
        0.043681715,
        -0.028183728,
        0.06203121,
        -0.0939594,
        0.033126526,
        0.063476264,
        -0.06942514,
        -0.006183725,
        -0.027545625,
        -0.16324167,
        -0.05970116,
        0.067197815,
        -0.050732773,
        -0.0053660283,
        0.04187308,
        0.090436354,
        -0.058052137,
        -0.048917066,
        0.10945992,
        0.009751313,
        0.011511046,
        0.02137321,
        -0.038931232,
        0.07878307,
        -0.0101697715,
        -0.023736302,
        -0.016212005,
        -0.00073608727,
        -0.08968139,
        0.041599438,
        -0.060304567,
        0.099050395,
        0.0780736,
        0.07943218,
        0.021101246,
        -0.054159828,
        0.06453427,
        -0.031191213,
        -0.06466519,
        -0.03471565,
        0.03010988,
        0.07100101,
        0.04052347,
        -0.029754734,
        -0.06780622,
        -0.5744202,
        0.01089918,
        0.043964434,
        -0.035182938,
        -0.14208865,
        0.07009296,
        -0.011221026,
        0.037264924,
        0.007642067,
        -0.05837548,
        0.07036531,
        -0.036710653,
        0.03018701,
        -0.025666421,
        -0.008053111,
        0.12001596,
        0.025658228
      ]
    },
    {
      "id": "worldsfair-speaker-261",
      "conference": "worldsfair",
      "name": "Serena Ge",
      "role": "Co-Founder & CEO",
      "company": "Datacurve",
      "twitter": "https://x.com/serenaa_ge",
      "talks": [
        {
          "title": "DeepSWE: expert code datasets",
          "description": "DeepSWE and the data/eval layer behind coding agents; why curated expert code datasets matter for reliable agent performance.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.010214863,
        0.08805142,
        -0.015279355,
        0.026420247,
        0.029368076,
        -0.045538582,
        -0.05771163,
        -0.020709839,
        0.012980346,
        -0.18613914,
        -0.06289402,
        -0.011227809,
        -0.0028409383,
        -0.0161115,
        -0.013456133,
        -0.031353284,
        0.06277533,
        -0.07656971,
        -0.08023393,
        -0.093836166,
        -0.0025320824,
        -0.066348605,
        0.0073742075,
        0.039316393,
        -0.049184993,
        0.14771506,
        -0.0028646502,
        0.087265566,
        -0.10423082,
        0.46116245,
        -0.0249187,
        0.029918898,
        0.072312206,
        0.018543478,
        0.06988532,
        -0.055940285,
        0.022271473,
        -0.05908005,
        0.027924173,
        0.07108787,
        -0.037197378,
        0.019070309,
        -0.025055656,
        0.047290795,
        -0.03903836,
        -0.03627378,
        -0.05876699,
        0.021719463,
        0.0028183642,
        -0.008892199,
        -0.04602216,
        -0.023370666,
        0.03355541,
        -0.000045801164,
        0.030302411,
        0.02402596,
        0.0029428175,
        -0.11375831,
        0.060016703,
        -0.0048699374,
        -0.19876204,
        0.026664779,
        0.03586447,
        -0.008637405,
        -0.03875594,
        -0.08333168,
        0.008707383,
        -0.0198561,
        0.0009253234,
        -0.08376862,
        0.041636094,
        -0.021533014,
        -0.037270226,
        0.07094397,
        -0.1037399,
        -0.17954102,
        -0.006586999,
        0.07265251,
        -0.06619808,
        0.01797516,
        -0.022188248,
        0.009592939,
        -0.06761357,
        -0.030256132,
        0.05236809,
        0.031147528,
        -0.039438903,
        0.025593733,
        -0.024034295,
        0.05125131,
        0.048386768,
        -0.03704762,
        -0.0660128,
        -0.013388513,
        -0.07153263,
        0.025049621,
        -0.0963154,
        -0.012012773,
        -0.0049386523,
        -0.013521954,
        -0.022036059,
        0.015290289,
        0.045609668,
        -0.013399476,
        -0.018717237,
        0.04131174,
        -0.008253928,
        -0.006160449,
        0.06144923,
        -0.021588309,
        -0.049726907,
        -0.6042474,
        -0.018568378,
        -0.03082891,
        -0.048548978,
        -0.15617184,
        -0.023822261,
        0.0036318211,
        0.035790406,
        0.076504074,
        -0.087562874,
        0.105595,
        -0.047138616,
        0.014045242,
        -0.06053249,
        0.017405283,
        0.05849039,
        0.012457786
      ]
    },
    {
      "id": "worldsfair-speaker-262",
      "conference": "worldsfair",
      "name": "Stephen Chin",
      "role": "VP of Developer Relations",
      "company": "Neo4j",
      "twitter": "https://x.com/steveonjava",
      "talks": [
        {
          "title": "CrabRAG: Why Automated Assistants Need Graph Memory, Not More Tokens",
          "description": "Autonomous assistants are easy to demo and hard to make reliable. The problem is usually not tool access. It is memory. Most assistant architectures still treat memory as a chat log plus vector retrieval. That is fine for document question answering, but it breaks down when the assistant must connect conversations, people, tools, and decisions across multiple tool iterations. For an AI engineer, a single request can depend on a Slack thread, a GitHub PR, a failed CI run, a calendar event, and prior operating preferences or constraints. These are not isolated pieces of context. They form a connected state that changes as work progresses and context grows. In this talk, I’ll show why knowledge graphs, context graphs, and GraphRAG provide a better foundation for OpenClaw-style assistants. Knowledge graphs capture durable entities and relationships. Context graphs capture the operational layer assistants usually lose, including actions, decision traces, provenance, and recency. GraphRAG turns that structure into task-time context by combining graph traversal, semantic retrieval, and tool use. Attendees will leave with practical patterns for schema design, retrieval routing, and evaluation, plus a concrete blueprint for assistants that remember more than the last prompt and retrieve more than the nearest chunk.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.03008467,
        0.071630195,
        -0.05999182,
        0.041894853,
        0.00524627,
        -0.01008321,
        -0.072768904,
        0.006028206,
        0.068725534,
        -0.20293991,
        -0.064972736,
        -0.085642435,
        0.012647775,
        -0.04771725,
        -0.064520635,
        0.017972069,
        0.13662694,
        -0.055579983,
        -0.105208255,
        -0.05777408,
        0.013642208,
        -0.054587908,
        -0.0059094024,
        0.068376765,
        -0.07897317,
        0.16555867,
        0.05336081,
        0.046141762,
        -0.05396222,
        0.45887315,
        -0.098383255,
        0.050302826,
        0.020229667,
        -0.022167271,
        0.05080679,
        -0.05889406,
        -0.021622792,
        -0.00921156,
        -0.013796873,
        0.014524698,
        -0.004318904,
        0.0029622228,
        -0.0031514445,
        0.02755953,
        -0.046204735,
        -0.04590906,
        -0.056877352,
        -0.011093334,
        0.015568151,
        -0.03508544,
        -0.048250485,
        -0.0026561893,
        0.03015503,
        0.0044258987,
        0.02083067,
        0.015270858,
        0.087265104,
        -0.0932612,
        0.11111198,
        -0.062396925,
        -0.15221635,
        -0.042982154,
        -0.015695408,
        -0.0032096675,
        -0.034765884,
        -0.061949853,
        0.056137927,
        -0.03529451,
        0.00858721,
        -0.057835314,
        0.0038473697,
        0.033709195,
        -0.005509252,
        0.032960385,
        -0.0059811953,
        -0.09809817,
        -0.05935789,
        0.062283717,
        -0.025138138,
        0.005164835,
        0.0226559,
        0.022224665,
        -0.0436549,
        -0.102991655,
        0.035239056,
        -0.04901864,
        -0.05212689,
        0.066285856,
        -0.047751207,
        0.06406923,
        0.015098951,
        -0.0519849,
        -0.028800873,
        0.0066314796,
        -0.08033929,
        0.047673497,
        -0.017548095,
        0.036927093,
        -0.019738644,
        0.011315929,
        -0.015845058,
        -0.03851259,
        0.013752146,
        -0.089399174,
        -0.020400245,
        -0.0702076,
        0.015562193,
        0.012389144,
        0.06920266,
        0.0596545,
        -0.0058915745,
        -0.60114896,
        -0.007878399,
        -0.022718709,
        -0.0109886555,
        -0.1525315,
        0.048047297,
        -0.055758033,
        0.07159649,
        0.101310946,
        -0.013975594,
        0.1265119,
        0.03852459,
        -0.0064084297,
        -0.015929956,
        0.044113897,
        0.050268847,
        -0.002609537
      ]
    },
    {
      "id": "worldsfair-speaker-263",
      "conference": "worldsfair",
      "name": "Everett Berry",
      "role": "",
      "company": "Clay",
      "twitter": "",
      "talks": [
        {
          "title": "Clay: AI in GTM",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.014328684,
        0.08845663,
        0.059402857,
        -0.027420202,
        0.011710468,
        0.018234717,
        -0.159367,
        0.025050225,
        -0.021380492,
        -0.15092266,
        -0.038875755,
        -0.051580742,
        0.023560805,
        -0.052075222,
        -0.029906765,
        -0.010634383,
        0.057179123,
        -0.0366788,
        -0.06080738,
        -0.024118505,
        -0.101425536,
        -0.089364804,
        -0.0065488648,
        0.01600958,
        -0.10094541,
        0.12603587,
        -0.035410166,
        0.08311942,
        -0.07376414,
        0.471792,
        -0.06119606,
        0.0001809922,
        -0.002030545,
        0.071472764,
        0.0074694534,
        -0.05261064,
        0.015902216,
        -0.009100485,
        0.007716404,
        0.01835311,
        -0.013090565,
        0.06867124,
        -0.043616686,
        0.020110294,
        0.028832916,
        0.0013330982,
        -0.0020864233,
        -0.07284417,
        -0.014650566,
        -0.0332972,
        -0.014641744,
        0.0044041197,
        0.002802998,
        -0.04851244,
        0.07195759,
        0.061600413,
        0.03678383,
        -0.041592237,
        0.0864111,
        -0.007634759,
        -0.14457417,
        0.0041945614,
        0.00811285,
        -0.00502972,
        0.019094786,
        -0.01110848,
        0.06918032,
        -0.058236703,
        -0.037636094,
        -0.09483115,
        -0.027104786,
        0.086942725,
        -0.051508892,
        0.010177536,
        -0.07669349,
        -0.022956494,
        -0.005940233,
        0.05251268,
        0.0069322297,
        -0.05160243,
        0.076329105,
        0.04917626,
        -0.05843436,
        -0.06740736,
        0.020720193,
        -0.061282158,
        -0.054812625,
        0.019467425,
        -0.020297734,
        0.07869905,
        0.013419901,
        -0.034073513,
        0.0023546743,
        0.019587424,
        -0.03394295,
        0.014717439,
        -0.05379559,
        0.04560898,
        -0.025271373,
        -0.00762616,
        -0.062091216,
        -0.012266801,
        0.0701109,
        -0.048849568,
        -0.086096466,
        -0.020981759,
        0.005914929,
        0.03157111,
        0.082246885,
        0.010497798,
        -0.038591783,
        -0.63918674,
        -0.03521679,
        -0.014793867,
        0.014788023,
        -0.11715387,
        -0.048744343,
        0.026812345,
        0.101929545,
        0.07666107,
        -0.04091372,
        0.0020105226,
        0.003913091,
        0.03763663,
        0.027865073,
        0.10001486,
        0.04868288,
        0.014016438
      ]
    },
    {
      "id": "worldsfair-speaker-264",
      "conference": "worldsfair",
      "name": "Chaitanya Asawa",
      "role": "Engineering, Clinical Decision Support",
      "company": "Abridge",
      "twitter": "",
      "talks": [
        {
          "title": "From Ambient Documentation to Clinical Intelligence",
          "description": "A practical session on how healthcare AI moves beyond ambient note generation into context-aware clinical decision support. The talk would cover grounding outputs in the patient encounter, surfacing evidence with citations inside clinician workflows, preserving clinician agency, and building rigorous evals for safety and trust in live healthcare environments.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.013112536,
        0.08193287,
        -0.036460146,
        0.006651075,
        0.0525794,
        -0.005902764,
        -0.09116455,
        -0.013646907,
        0.05394106,
        -0.17390214,
        -0.05720908,
        -0.027916338,
        0.027084574,
        -0.073315084,
        -0.019099556,
        -0.012332244,
        0.07715225,
        -0.057847705,
        -0.03326956,
        -0.0364058,
        -0.052739806,
        -0.038794193,
        -0.029481865,
        0.076082736,
        -0.06747431,
        0.11649032,
        0.040744133,
        0.092062354,
        -0.1255581,
        0.43006662,
        -0.17997089,
        -0.03348334,
        -0.0075242626,
        0.008962232,
        0.07526628,
        -0.03039347,
        0.04008474,
        -0.04909552,
        0.034206577,
        0.03460115,
        0.02257249,
        0.038980912,
        -0.039693747,
        0.033275902,
        0.042488657,
        -0.008772759,
        -0.0484743,
        0.027358344,
        0.0007319702,
        0.00907645,
        -0.07828119,
        -0.051241938,
        0.05300887,
        0.004099546,
        -0.0041631954,
        0.038269438,
        -0.03504237,
        -0.06369143,
        0.1150574,
        -0.0076351454,
        -0.06195942,
        -0.005404377,
        0.021386316,
        -0.035283174,
        0.036183994,
        -0.09466901,
        0.100895576,
        -0.009130422,
        -0.048266023,
        -0.015061426,
        0.014235058,
        -0.0066953897,
        -0.04908476,
        -0.015553664,
        -0.045267776,
        -0.09480027,
        -0.04492349,
        0.1101138,
        -0.013726913,
        0.083291814,
        0.004358607,
        0.046018753,
        -0.06717845,
        -0.06981038,
        0.027202165,
        0.067568816,
        -0.044232786,
        0.028723683,
        0.016853217,
        0.021433821,
        0.017405929,
        -0.043206144,
        -0.04375283,
        -0.071398325,
        -0.028870698,
        -0.00168407,
        0.019727513,
        0.026579015,
        -0.053806484,
        0.023136845,
        0.0028244741,
        -0.04261839,
        0.07065522,
        0.007896934,
        0.031535428,
        0.05949407,
        0.009909825,
        0.03293977,
        0.06323203,
        0.013307852,
        -0.031957265,
        -0.6420893,
        0.04551518,
        0.0011135328,
        -0.09274457,
        -0.16079162,
        0.028583217,
        0.016643288,
        0.04599886,
        0.12651291,
        -0.047536097,
        0.057166867,
        0.020356936,
        -0.06137712,
        -0.013654095,
        -0.0093795415,
        0.04783856,
        0.008316477
      ]
    },
    {
      "id": "worldsfair-speaker-265",
      "conference": "worldsfair",
      "name": "Nishant Gupta",
      "role": "Staff Software Engineer and Researcher",
      "company": "Meta",
      "twitter": "",
      "talks": [
        {
          "title": "Operating Distributed Inference Systems at Scale",
          "description": "Inference has rapidly become one of the most important infrastructure problems in modern computing. As AI systems evolve into autonomous agents with persistent memory, tool usage, and multi-step reasoning, traditional inference architectures struggle under growing demands for latency, throughput, cost efficiency, and reliability. In this talk, I’ll share lessons from building large-scale elastic compute and AI infrastructure systems powering production workloads. We’ll explore the modern inference stack and the architectural patterns emerging to support next-generation agentic AI systems. Topics include distributed inference architectures for large-scale AI systems, GPU scheduling and elastic compute for inference workloads, multi-tenant inference infrastructure, caching, batching, latency optimization strategies, reliability and fault isolation for inference systems, observability and control loops for AI serving platforms, balancing cost, throughput, and user experience, and why inference is becoming an infrastructure orchestration problem. Attendees will gain practical insights into designing scalable, resilient, and cost-efficient inference platforms for modern AI workloads.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.045581862,
        0.11815195,
        -0.03653254,
        0.06126526,
        0.008586574,
        -0.09321608,
        -0.05632821,
        -0.052890938,
        0.051373716,
        -0.24379902,
        -0.018593205,
        -0.065768,
        0.0036126971,
        -0.051945254,
        -0.04922661,
        0.025259167,
        0.11662847,
        -0.03645276,
        -0.08543763,
        -0.05183072,
        -0.01291571,
        -0.119461335,
        -0.012595047,
        0.06381723,
        0.02138251,
        0.075484745,
        0.036356434,
        0.03511495,
        -0.12518048,
        0.41175136,
        -0.10695111,
        -0.01821042,
        0.027320664,
        -0.058242366,
        0.0567432,
        -0.053473327,
        -0.04943263,
        0.021398287,
        0.031081362,
        0.06518832,
        -0.0061448626,
        0.08300454,
        -0.057840932,
        0.033976886,
        0.007702682,
        0.0017415915,
        -0.12611474,
        0.027816396,
        -0.0044318917,
        -0.02546812,
        -0.05857663,
        -0.0028298127,
        0.047181647,
        0.022805413,
        -0.03262037,
        0.055535987,
        -0.040902812,
        -0.106863886,
        0.13901377,
        -0.046926472,
        -0.13176236,
        -0.012770354,
        -0.00958662,
        -0.0074836276,
        0.012131542,
        -0.10702694,
        0.038711693,
        -0.094144456,
        -0.045716934,
        0.024537005,
        0.008324583,
        -0.043238193,
        -0.040530335,
        0.019892745,
        -0.022985918,
        -0.093833596,
        -0.013497583,
        0.06105666,
        -0.017731225,
        -0.00047743125,
        0.02951187,
        0.06105608,
        -0.0034591958,
        0.022833874,
        0.10282628,
        -0.032305434,
        0.017915223,
        0.008208973,
        -0.0418538,
        0.09908363,
        0.074454784,
        -0.07418057,
        -0.05005376,
        0.0040510776,
        -0.0354553,
        0.0073085106,
        -0.0839326,
        0.06165752,
        0.017596062,
        0.06700097,
        0.04736311,
        -0.0043840394,
        0.048068106,
        -0.017471656,
        -0.031702343,
        -0.047874518,
        0.026924932,
        -0.0034568491,
        0.08794218,
        -0.014704254,
        -0.102579825,
        -0.5973243,
        0.012914337,
        0.020499807,
        -0.05834844,
        -0.15225357,
        0.03919762,
        -0.054556653,
        0.030839311,
        0.03184881,
        -0.042458456,
        0.083789274,
        -0.007713355,
        0.002677761,
        0.029481,
        0.05405688,
        0.024441116,
        -0.0038334925
      ]
    },
    {
      "id": "worldsfair-speaker-266",
      "conference": "worldsfair",
      "name": "Jennifer Lee",
      "role": "Product Manager",
      "company": "Stripe",
      "twitter": "",
      "talks": [
        {
          "title": "Building safe payment infrastructure for the autonmous economy",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.022226235,
        -0.0383268,
        -0.046843734,
        0.020444484,
        0.0018520541,
        -0.05729505,
        -0.097469226,
        0.047182377,
        0.0881759,
        -0.23129907,
        -0.041745573,
        -0.053440258,
        -0.048044026,
        0.019372867,
        -0.04471477,
        0.09265866,
        0.05837338,
        -0.08453653,
        -0.009410338,
        -0.040092576,
        -0.054653727,
        -0.086017504,
        0.05250832,
        -0.0020291996,
        -0.016738186,
        0.056211125,
        0.037290033,
        0.03969378,
        -0.082406,
        0.48650998,
        -0.031821586,
        0.041868716,
        -0.013102223,
        -0.0372197,
        0.06446049,
        0.0050313417,
        0.022237904,
        -0.00882185,
        0.016428685,
        0.028350247,
        -0.020272376,
        0.018955301,
        -0.06963154,
        0.0011640899,
        0.009925336,
        -0.042520177,
        -0.13123457,
        -0.030216344,
        0.052306145,
        0.008321442,
        -0.074900545,
        -0.019071624,
        -0.013078202,
        -0.012824618,
        0.045163833,
        -0.0054493435,
        0.015464798,
        -0.07998113,
        0.113080315,
        -0.054675985,
        -0.11069414,
        0.03913696,
        0.010881915,
        0.01779299,
        0.024242165,
        -0.07572369,
        0.024607485,
        0.0072457334,
        0.031745553,
        -0.028279712,
        -0.010297058,
        0.013289297,
        -0.027276844,
        0.09567874,
        -0.09941387,
        -0.09166538,
        0.00262777,
        0.059817955,
        -0.034105606,
        0.08019233,
        0.06159502,
        0.01104453,
        -0.03184182,
        -0.06916262,
        -0.022881899,
        0.0050022877,
        0.013244392,
        0.060117558,
        -0.06528682,
        0.10029011,
        -0.063440554,
        -0.006583635,
        -0.028297845,
        -0.012138439,
        -0.034368258,
        0.1034693,
        -0.077917226,
        0.08803447,
        0.057944212,
        0.052408624,
        -0.09209424,
        -0.07231361,
        -0.03321263,
        -0.015788881,
        -0.08715766,
        -0.045266856,
        -0.027875941,
        0.0010246866,
        0.053181402,
        0.028509367,
        0.0012611457,
        -0.58502597,
        0.021158695,
        -0.028556945,
        -0.017135672,
        -0.127374,
        0.00500402,
        -0.051217005,
        0.009082258,
        0.14179176,
        -0.034610808,
        0.064410985,
        -0.078433976,
        0.01834812,
        -0.006385059,
        0.052913424,
        0.044430487,
        0.0084969355
      ]
    },
    {
      "id": "worldsfair-speaker-267",
      "conference": "worldsfair",
      "name": "Thorsten Hans",
      "role": "",
      "company": "Akamai",
      "twitter": "",
      "talks": [
        {
          "title": "Edge-Native AI: Building Ultra-Fast Agents and MCP Servers with Spin",
          "description": "Centralized AI is slow; Edge-native AI is the revolution. Thorsten Hans demonstrates how to build intelligent agents and Model Context Protocol (MCP) servers that run at the speed of light. Using Spin and WebAssembly, we'll bypass the \"cloud tax\" of high latency and cold starts. Discover how to ship AI-driven features that live closer to your users, ensuring sub-millisecond responsiveness and enhanced privacy. Stop waiting for the origin it's time to bring the brain to the edge and master the stack that powers the next generation of intelligent, distributed applications.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.038966574,
        0.10131378,
        0.039519016,
        -0.0044383607,
        0.05655347,
        -0.041183256,
        -0.08809705,
        -0.036661368,
        0.0435341,
        -0.20234136,
        -0.040600408,
        0.02978718,
        0.012705804,
        -0.01848984,
        -0.08851605,
        -0.0070333094,
        0.12429584,
        -0.041217852,
        -0.04649061,
        -0.060671985,
        0.0002583658,
        -0.056581937,
        -0.029941492,
        0.04061436,
        0.0027825055,
        0.11587612,
        -0.053996578,
        0.028530398,
        -0.12624055,
        0.4985811,
        -0.07826806,
        0.029853823,
        0.011616757,
        -0.016172687,
        0.052069996,
        -0.027786287,
        -0.054065365,
        -0.03431475,
        0.02973649,
        0.044358782,
        -0.039232966,
        0.09445197,
        -0.07294536,
        0.018160746,
        -0.053968012,
        -0.013068529,
        -0.03654187,
        0.021673419,
        -0.008355313,
        -0.01074958,
        -0.041230157,
        0.063446835,
        0.06140154,
        -0.019528326,
        0.07044501,
        0.013008893,
        0.08029722,
        -0.05655971,
        0.08873792,
        -0.0689171,
        -0.059851363,
        -0.02712753,
        -0.022198377,
        0.017020734,
        -0.025111387,
        -0.10050219,
        0.04758767,
        -0.04425908,
        -0.005799165,
        -0.047407053,
        -0.0140184155,
        0.017885895,
        -0.043525446,
        0.05294857,
        -0.06286104,
        -0.0476437,
        -0.06594045,
        0.027984012,
        -0.038716763,
        -0.034065012,
        -0.021757491,
        0.023974217,
        -0.01870975,
        -0.052703794,
        0.059745975,
        -0.0124735255,
        -0.08386046,
        -0.035749473,
        -0.020015366,
        0.045505296,
        0.02113525,
        0.025030069,
        -0.02947116,
        0.022194317,
        -0.03753269,
        -0.00402699,
        -0.051770516,
        -0.025159374,
        0.018987298,
        -0.0059480476,
        -0.042097133,
        -0.05212841,
        0.016710045,
        -0.012912195,
        -0.027681613,
        -0.057677254,
        0.026745412,
        -0.0117876455,
        -0.014745968,
        -0.0080261715,
        0.005183799,
        -0.62916464,
        0.011739875,
        -0.041978974,
        -0.024821188,
        -0.15891877,
        0.03229811,
        -0.02901021,
        0.08228199,
        0.0799901,
        -0.03372876,
        0.069635846,
        -0.059988,
        -0.027515259,
        0.050139204,
        0.047568347,
        -0.007479079,
        0.016874582
      ]
    },
    {
      "id": "worldsfair-speaker-268",
      "conference": "worldsfair",
      "name": "Ryan Cooke",
      "role": "",
      "company": "WorkOS",
      "twitter": "",
      "talks": [
        {
          "title": "No, That's Not a Software Factory",
          "description": "Drop an agent in a sandbox, point it at your repo, watch it ship code. Whether you're buying from a vendor or building it yourself, everyone is following the same playbook. But a sandbox isn't a software factory. At WorkOS, we built Project Horizon, and it taught us that infrastructure is only the first challenge. The unlock is encoding how your org actually builds software: the way work gets planned, scoped, and verified, the conventions and judgment calls that define your engineering culture. Our product engineering process served as the blueprint for every agent workflow we built in Horizon.",
          "day": "Day 4 — Session Day 3",
          "time": "10:45am-11:05am",
          "type": "session"
        }
      ],
      "embedding": [
        0.04267652,
        0.097002424,
        0.018304655,
        -0.056511577,
        -0.006802107,
        -0.031629905,
        -0.0917493,
        -0.050109286,
        0.046473082,
        -0.258417,
        -0.015728345,
        -0.037646025,
        -0.009904036,
        -0.02051759,
        -0.049485512,
        0.015775613,
        0.072361305,
        -0.021873584,
        -0.02977989,
        -0.06684795,
        -0.07885194,
        -0.042712476,
        0.02776149,
        0.026711833,
        -0.050142203,
        0.12558271,
        0.017629484,
        0.040120564,
        -0.10040179,
        0.4513133,
        -0.07812822,
        0.0654479,
        -0.0015166282,
        -0.110245556,
        0.099094,
        -0.041403603,
        -0.085338056,
        0.050240714,
        0.049727924,
        0.045144916,
        -0.04976754,
        0.041038726,
        -0.070786,
        0.044210196,
        -0.044031832,
        -0.022605764,
        -0.052878376,
        -0.018336097,
        -0.012096724,
        0.0040113567,
        -0.070104666,
        0.024966927,
        0.04007296,
        0.0073509407,
        0.083988614,
        -0.013544897,
        0.00702634,
        -0.109265454,
        0.09249424,
        -0.044328235,
        -0.10059942,
        0.016270101,
        0.012033417,
        0.014809947,
        -0.029951176,
        -0.05953519,
        0.046552323,
        -0.06157871,
        -0.011505735,
        -0.08071291,
        0.02411974,
        0.044480197,
        -0.076203026,
        0.055122804,
        -0.1259795,
        -0.12847011,
        -0.018543923,
        0.067049764,
        -0.0036295743,
        0.0028477616,
        0.024304299,
        0.06323709,
        -0.0698006,
        -0.032718673,
        0.032366063,
        -0.0182321,
        -0.016731068,
        -0.039658573,
        -0.020689087,
        0.04046502,
        0.018983487,
        -0.05242055,
        0.022222187,
        0.03708556,
        -0.049813677,
        -0.03614263,
        -0.04036484,
        0.043412793,
        -0.045643907,
        0.026708962,
        -0.008935587,
        -0.0014184014,
        -0.039500795,
        0.015940277,
        -0.02802691,
        -0.04155762,
        0.0047492394,
        -0.039035127,
        0.050917353,
        -0.00852776,
        0.029409451,
        -0.6118922,
        -0.06178178,
        0.015412336,
        -0.03368899,
        -0.1529442,
        0.020963848,
        0.042498652,
        0.039078977,
        0.07540918,
        -0.025112081,
        0.07942046,
        -0.039326165,
        -0.0030273711,
        0.024754325,
        0.06489419,
        0.00021575957,
        0.008193486
      ]
    },
    {
      "id": "worldsfair-speaker-269",
      "conference": "worldsfair",
      "name": "Michael Grinich",
      "role": "Founder & CEO",
      "company": "WorkOS",
      "twitter": "https://x.com/grinich",
      "talks": [
        {
          "title": "Auth for Agents: Unblock Autonomous AI with auth.md",
          "description": "AI agents are ready to act on users' behalf, but legacy auth flows were built for humans, not agents. This session introduces auth.md, an open protocol that lets agents register and authenticate users without sign-up forms, and shares what early implementers have learned since launch. Learn about the new protocol that Cloudflare, Firecrawl, Cogny, and monday.com are adopting to power agent registration — authenticating agents without sign-up forms.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.025382834,
        0.027047167,
        -0.004150649,
        -0.054469798,
        -0.015874133,
        -0.013297072,
        -0.060308393,
        -0.04161757,
        0.07798673,
        -0.22801462,
        -0.031281304,
        -0.05590515,
        0.018926175,
        -0.012709616,
        -0.011488705,
        0.0035681794,
        0.048618026,
        -0.07049883,
        -0.06208591,
        -0.10109682,
        -0.06612141,
        -0.052828576,
        0.04195481,
        0.01434437,
        -0.054895047,
        0.115076534,
        0.050243128,
        0.06793176,
        -0.11050702,
        0.43397346,
        -0.09695221,
        0.005293289,
        -0.010239948,
        -0.018468032,
        0.06390303,
        -0.071667336,
        -0.032687407,
        0.013777837,
        0.028343376,
        0.024822315,
        0.02050575,
        -0.007911338,
        -0.07463704,
        -0.029776745,
        0.039490975,
        -0.061010286,
        -0.056672618,
        0.010342207,
        0.01264928,
        -0.002873861,
        -0.07399122,
        0.045510184,
        0.07744254,
        0.05248172,
        0.04865029,
        0.021808544,
        0.027018271,
        -0.057939354,
        0.071201354,
        -0.09344806,
        -0.09256221,
        0.0017078968,
        -0.0008346037,
        0.033715762,
        -0.00051494327,
        -0.06771432,
        0.066685654,
        -0.0037947453,
        0.03231746,
        -0.09275653,
        0.048395246,
        0.08010845,
        -0.06751694,
        0.043470502,
        -0.055789128,
        -0.17440507,
        0.0062589697,
        0.07752488,
        -0.043865394,
        0.004091808,
        -0.009260803,
        0.065072075,
        -0.053416457,
        -0.0766484,
        -0.006449907,
        -0.057490896,
        -0.041005183,
        0.035602365,
        0.006103424,
        -0.000828351,
        -0.011450004,
        -0.039084498,
        -0.03990416,
        0.046601214,
        -0.034505583,
        -0.031551108,
        -0.040888708,
        0.00030254322,
        -0.016680153,
        -0.017646722,
        0.02696832,
        0.004422976,
        0.10284331,
        0.032850854,
        -0.014657611,
        0.004656491,
        0.07089504,
        0.011220937,
        0.049582224,
        0.00936189,
        -0.018140728,
        -0.64142483,
        -0.015013473,
        0.041790377,
        -0.012493861,
        -0.17127219,
        0.008125345,
        0.018312344,
        0.029381452,
        0.07893897,
        -0.05406897,
        0.068220384,
        -0.04698429,
        -0.004572356,
        0.0077987113,
        0.03119803,
        0.04171521,
        0.00007203743
      ]
    },
    {
      "id": "worldsfair-speaker-270",
      "conference": "worldsfair",
      "name": "James Russo",
      "role": "Senior Software Engineer, project lead",
      "company": "HeyGen",
      "twitter": "https://x.com/Rames_Jusso",
      "talks": [
        {
          "title": "HTML Is All Agents Need",
          "description": "AI agents compose videos by writing HTML, CSS, and JS.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.03005235,
        0.06450019,
        0.06914251,
        0.036340423,
        -0.0043943245,
        -0.026831426,
        -0.09383865,
        0.012039487,
        -0.011404819,
        -0.2260347,
        -0.030037237,
        -0.03095364,
        0.03758372,
        -0.019882571,
        -0.0065411595,
        0.021863958,
        0.11615964,
        -0.092247695,
        0.003488899,
        -0.08901481,
        -0.04718324,
        -0.11845394,
        -0.037251282,
        0.04915522,
        -0.047093328,
        0.109728836,
        -0.009363895,
        0.04985996,
        -0.06352979,
        0.45503893,
        -0.09076633,
        0.05350999,
        0.007471406,
        -0.0460732,
        0.0369088,
        -0.07302025,
        -0.04528298,
        -0.06043198,
        -0.03656628,
        0.07274994,
        0.03131231,
        0.049248535,
        0.004150011,
        0.01957668,
        0.040895894,
        0.0018963686,
        -0.050073892,
        0.009710815,
        0.056213908,
        -0.054136608,
        -0.06793648,
        -0.011480463,
        0.049672443,
        -0.06991698,
        -0.0143525945,
        -0.05909177,
        0.04130369,
        -0.05140658,
        0.04277662,
        -0.01086208,
        -0.04134526,
        -0.056846283,
        0.07485274,
        -0.036879066,
        0.042687554,
        -0.048274927,
        0.056246456,
        -0.028831333,
        0.014043935,
        -0.07834107,
        -0.07475943,
        0.025498265,
        -0.06905896,
        0.0499054,
        -0.12681998,
        -0.07510019,
        -0.095390566,
        0.014260759,
        -0.013091054,
        0.031344198,
        -0.009564485,
        0.056320153,
        -0.04516471,
        -0.04707022,
        0.019121233,
        0.011532042,
        -0.041100696,
        0.032198343,
        -0.020370869,
        0.055735912,
        0.046324812,
        -0.04212448,
        0.010885799,
        0.055512946,
        -0.036912028,
        0.006815773,
        -0.06337841,
        -0.028928379,
        0.01844499,
        0.01373438,
        -0.0743099,
        -0.04479668,
        -0.06302286,
        -0.0069195842,
        0.016514333,
        -0.048548114,
        0.021004146,
        -0.0011691137,
        0.12894674,
        0.013304544,
        -0.010785367,
        -0.62168723,
        -0.053176396,
        -0.015648996,
        -0.018948214,
        -0.13212757,
        0.029741399,
        0.0047609233,
        0.021576243,
        0.08310688,
        -0.017475447,
        0.13441816,
        0.0067311237,
        -0.03869071,
        -0.038262907,
        0.04657889,
        0.07704685,
        -0.017156549
      ]
    },
    {
      "id": "worldsfair-speaker-271",
      "conference": "worldsfair",
      "name": "Harshal Bhangale",
      "role": "Staff Software Engineer",
      "company": "Circle",
      "twitter": "",
      "talks": [
        {
          "title": "Why Your AI Agent Needs a Wallet: Agentic commerce on Arc with USDC and Nanopay",
          "description": "AI agents can reason, plan, call tools, and write code. But the moment one needs paid data, an API call, or another agent's service, it hits a human wall: accounts, API keys, credit cards, checkout flows. It stalls and asks you to step in. It can't pay. We'll run the same real task through two agents, one without a wallet and one with. The first stalls. The second, handed a Circle agent wallet through the Circle CLI, discovers services, pays per request over x402 in USDC, and finishes on its own, inside spending limits you set. The next leap in agents isn't only better models or more tools. It's economic agency: holding programmable money and transacting at machine speed. We'll show how it works on Arc, where USDC is the gas, finality is sub-second, and gasless nanopayments settle in batches through Circle Gateway, so paying a fraction of a cent per request is actually practical.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.038913034,
        0.046445336,
        0.028543016,
        0.066171296,
        0.045281734,
        -0.011243668,
        -0.10099762,
        0.014163049,
        -0.015303967,
        -0.22425494,
        -0.06873637,
        -0.020451082,
        0.025674414,
        -0.05548608,
        -0.0700087,
        -0.004099111,
        0.09693055,
        -0.082626976,
        -0.053858627,
        -0.09152392,
        -0.09105016,
        -0.0946666,
        -0.020666363,
        0.033439018,
        -0.06462714,
        0.11449614,
        0.0045196465,
        0.077180654,
        -0.09213164,
        0.48769808,
        -0.085749276,
        0.022586115,
        0.022331428,
        -0.059977863,
        0.083986096,
        -0.08847508,
        -0.029481221,
        -0.03066324,
        -0.0031237355,
        0.029869536,
        0.0021058568,
        0.011682452,
        -0.07069461,
        0.055085413,
        -0.021783628,
        -0.003786736,
        -0.07401848,
        0.010298627,
        0.014487471,
        -0.058883,
        -0.07887901,
        0.018538304,
        0.055465,
        -0.0017847453,
        0.07150712,
        0.023605358,
        0.0096341,
        -0.047025062,
        0.105952844,
        -0.06589152,
        -0.08022857,
        -0.0024000872,
        0.022529518,
        0.045380678,
        0.040850848,
        -0.036937237,
        0.08084343,
        -0.06003525,
        0.01026946,
        -0.010444452,
        -0.025895588,
        0.0032712978,
        -0.05113756,
        0.093277685,
        -0.05390653,
        -0.08991448,
        -0.06942111,
        0.035553895,
        0.00032120687,
        0.0105495015,
        -0.015973141,
        0.007938937,
        -0.05200424,
        -0.023455972,
        -0.0041514235,
        -0.029270858,
        -0.039562687,
        0.0291726,
        0.018562507,
        0.07619042,
        -0.028659273,
        -0.06685228,
        -0.06883102,
        -0.020469332,
        -0.06341621,
        -0.0007752102,
        -0.07943243,
        0.04117785,
        -0.05556252,
        0.059311867,
        -0.035476733,
        -0.048087735,
        0.003004527,
        -0.027720032,
        -0.01941379,
        -0.056015685,
        0.030148238,
        -0.036223475,
        0.011895356,
        -0.014981142,
        -0.00035391492,
        -0.5945521,
        0.004444828,
        -0.03405821,
        -0.05580316,
        -0.15136366,
        0.043730173,
        0.007206676,
        0.01483343,
        0.10271373,
        -0.07440307,
        0.076051325,
        -0.045875557,
        0.00007707992,
        -0.013002577,
        0.019435676,
        0.07435657,
        0.024887132
      ]
    },
    {
      "id": "worldsfair-speaker-272",
      "conference": "worldsfair",
      "name": "Udi Menkes",
      "role": "Principal AI Product Manager",
      "company": "Intuit",
      "twitter": "",
      "talks": [
        {
          "title": "AI Doesn't Understand Money (Yet)",
          "description": "Ask any LLM a financial question about your business. You'll get a fluent, confident, generic answer — one that doesn't truly know your business, or what happened when businesses like yours made that same decision. We build financial AI at Intuit serving 100M+ customers. Our custom LLMs outperform general-purpose models on accuracy while cutting latency in half. But that's the foundation, not the destination. I'll cover where financial intelligence goes when AI stops reporting what happened and starts helping you decide what to do next (and does it for you).",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.03687139,
        0.047916867,
        -0.012471433,
        0.04080778,
        -0.0009462534,
        -0.0017642081,
        -0.09502545,
        0.03586013,
        0.04730543,
        -0.24963698,
        0.022967603,
        -0.018421765,
        -0.031295102,
        -0.044814024,
        -0.06855885,
        0.04897996,
        0.13243036,
        -0.041090023,
        -0.07429745,
        -0.13569105,
        -0.045470133,
        -0.04347241,
        -0.01198337,
        0.13842766,
        -0.083058506,
        0.11476353,
        0.10484586,
        0.015299732,
        -0.10163672,
        0.4478549,
        -0.10626107,
        0.0258088,
        -0.04526543,
        0.005013394,
        0.02492803,
        -0.028064596,
        -0.060585726,
        -0.017395863,
        -0.030228715,
        0.089076795,
        0.0044575967,
        0.044639684,
        -0.021642363,
        0.03275729,
        0.022294655,
        -0.07830655,
        -0.07157169,
        -0.0045797424,
        0.022001758,
        -0.030027049,
        -0.0077891503,
        0.010520618,
        0.03357539,
        -0.003231778,
        0.045641076,
        -0.005814129,
        0.036889542,
        -0.08828797,
        0.1192499,
        -0.04492,
        -0.066619515,
        0.060574125,
        -0.048742257,
        0.03102109,
        0.029096954,
        -0.07564031,
        0.02535814,
        -0.033029657,
        0.0025272667,
        0.026570901,
        -0.07348698,
        0.058003757,
        -0.08086287,
        0.062098194,
        -0.08807091,
        -0.15300573,
        -0.030895619,
        0.10315433,
        -0.01765131,
        -0.00015835685,
        0.027260475,
        0.060218666,
        -0.05045007,
        0.005977006,
        0.053420585,
        -0.055236194,
        -0.06641633,
        0.075228356,
        0.010781885,
        0.10832476,
        0.031670738,
        -0.0143541135,
        -0.031425316,
        -0.001582176,
        -0.06472418,
        0.030258086,
        -0.09065602,
        0.054916695,
        0.028527508,
        0.006679962,
        -0.0020876315,
        -0.035036698,
        0.055262867,
        0.011026419,
        -0.021505695,
        -0.017575547,
        -0.027020192,
        0.040818606,
        0.05999041,
        0.042031992,
        -0.007147534,
        -0.5803978,
        -0.005037255,
        0.018695693,
        -0.055108123,
        -0.1375762,
        0.09434853,
        -0.004127176,
        0.024288286,
        0.04492818,
        -0.031823114,
        0.012217665,
        0.0025295261,
        0.0043428736,
        -0.008207951,
        0.03858541,
        0.039549824,
        0.037197046
      ]
    },
    {
      "id": "worldsfair-speaker-273",
      "conference": "worldsfair",
      "name": "Frank Coyle",
      "role": "Computer Science Educator; Founder, The AI Edge",
      "company": "University California Berkeley",
      "twitter": "https://x.com/coyle_frankp",
      "talks": [
        {
          "title": "Anthropic's CCA Exam as a Field-Guide for Agentic Engineering",
          "description": "**Anthropic's CCA Exam: A Field-Guide for Agentic Engineering** The Claude Certified Architect (CCA) exam distills what Anthropic has learned from working with the AI companies shipping agents to production — the patterns that work, the anti-patterns that quietly burn tokens and trust, and the architectural decisions that separate demos from systems you'd stake a quarter on. This talk treats the exam as a field guide for agentic engineering, whether or not you ever sit for it. We'll walk through the five competency domains the exam tests — Agentic Architecture, Tool Design and MCP Integration, Claude Code, Prompt Engineering, and Context Management — with particular emphasis on multi-agent orchestration, subagent delegation, tool schema design, and lifecycle hooks. We'll then work through the six real-world scenarios the exam uses to probe judgment, each organized around an anti-pattern: the seductive-but-wrong move that looks reasonable until it costs you a production incident. Attendees leave with a working mental model of the agentic surface area and a checklist of the failure modes that matter most when moving from prototype to production. **Who should attend:** engineers and architects building agentic systems with Claude or other frontier models, technical leads evaluating agent designs, and developers considering the CCA credential.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        },
        {
          "title": "Why Agentic Systems Need Ontologies",
          "description": "Agentic systems fail in predictable ways: context degradation, brittle tool descriptions, fragile multi-agent handoffs, stop-reason confusion, and the ever-present temptation to fix reliability problems with more natural-language instructions. These anti-patterns aren't bugs to be patched turn by turn — they're symptoms of a missing architectural layer. LLMs reason probabilistically over domains they only partially understand, and no amount of prompt engineering fully closes that gap. This talk argues that the missing layer is an explicit ontology: a formal, shared map of the domain's concepts, relationships, and constraints. The pattern is not new — ontologies have driven commercial success in defense and intelligence systems for over a decade, where probabilistic models must operate over high-stakes enterprise data without drifting into nonsense. Graph databases like Neo4j and Amazon Neptune have made the underlying primitives widely accessible. We'll show how lightweight ontology constructs can surround an agentic system with enforceable logical constraints: typed entities and relationships that tools must respect, cardinality and domain restrictions that catch malformed tool calls before they execute, and a shared vocabulary that keeps coordinators and subagents talking about the same things. The session walks through several agentic applications — a multi-agent research workflow, a tool-heavy customer support agent, a coordinator-subagent delegation pattern — and shows in each case how an ontology layer addresses the kinds of anti-patterns catalogued in Anthropic's Claude Certified Architect exam. The result is a hybrid neurosymbolic architecture: probabilistic reasoning inside, logical guardrails outside. Who should attend: engineers building production agentic systems, architects evaluating reliability strategies beyond prompt engineering, and technical leads who suspect their agents need more structure than another system prompt can provide.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.11475823,
        0.029502407,
        0.014792411,
        0.011689624,
        0.07386042,
        -0.018067934,
        -0.11104088,
        -0.06994884,
        0.017519392,
        -0.20954996,
        -0.022892062,
        -0.012185431,
        0.01994006,
        -0.017821344,
        -0.03917077,
        -0.003218904,
        0.06518076,
        -0.0141329905,
        -0.02834949,
        -0.06854396,
        -0.11230494,
        -0.08212,
        -0.029203221,
        0.05918818,
        -0.07239524,
        0.1232461,
        0.046302687,
        0.1424219,
        -0.1363768,
        0.41012093,
        -0.07415613,
        0.04204119,
        0.062345855,
        0.000114346374,
        -0.015189787,
        -0.01917677,
        -0.045657285,
        -0.0071543,
        0.08095151,
        0.03451305,
        0.026280172,
        0.08684331,
        -0.047445606,
        0.006611004,
        -0.019828256,
        -0.017223995,
        -0.08839786,
        0.007714537,
        -0.032152347,
        -0.033258334,
        -0.055579484,
        -0.016995868,
        -0.000060025366,
        0.04564056,
        0.08052353,
        0.029827004,
        0.042659406,
        -0.09294863,
        0.105647184,
        -0.071198456,
        -0.06866502,
        -0.07858913,
        -0.034788802,
        -0.00137252,
        0.029636122,
        -0.07481804,
        0.06283816,
        -0.09273675,
        -0.00908246,
        -0.09105479,
        -0.0052642142,
        0.11481088,
        -0.06674396,
        0.028440055,
        -0.061504908,
        -0.11761341,
        -0.053640634,
        0.09638973,
        -0.031896774,
        0.016198892,
        -0.015979327,
        0.11716514,
        -0.060474038,
        -0.09962843,
        0.05669377,
        -0.04408961,
        -0.100445874,
        0.043042198,
        -0.04973511,
        0.028023086,
        0.053361345,
        -0.082271114,
        -0.058597308,
        -0.03961868,
        -0.06526068,
        -0.029386586,
        -0.025642632,
        0.08271721,
        -0.057372376,
        0.031235598,
        -0.034048155,
        -0.10389223,
        0.03461156,
        -0.039247148,
        0.0046267724,
        -0.11149468,
        -0.017278679,
        0.01897013,
        0.04531512,
        0.0022956375,
        -0.0556613,
        -0.5615224,
        0.032322727,
        -0.004044383,
        -0.0047122585,
        -0.08942204,
        0.12980993,
        0.016753763,
        0.0003038198,
        0.052082386,
        -0.020238508,
        0.06519333,
        0.05051345,
        -0.04300985,
        -0.056783203,
        -0.011203956,
        0.10797837,
        -0.0033205107
      ]
    },
    {
      "id": "worldsfair-speaker-274",
      "conference": "worldsfair",
      "name": "Yohei Nakajima",
      "role": "Managing Partner",
      "company": "BabyAGI/Untapped Capital",
      "twitter": "https://x.com/yoheinakajika",
      "talks": [
        {
          "title": "Active Graph Agent Runtime (BabyAGI 4)",
          "description": "Proposing a novel event-sourced graph runtime for building long-running auditable, agentic systems. Built on top of and combining various BabyAGI iterations and graph experiments (memory, code, logs) into a single primitive.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.010162537,
        0.07671086,
        0.013617465,
        0.025923563,
        -0.028683096,
        0.020496875,
        -0.06927206,
        -0.017830726,
        0.07133441,
        -0.24501939,
        -0.016617185,
        -0.03730949,
        -0.031619605,
        -0.01692227,
        -0.077652454,
        0.009905763,
        0.063739456,
        -0.04963502,
        -0.009755406,
        -0.08986155,
        -0.033757243,
        -0.07228378,
        -0.03197197,
        0.085814685,
        -0.06720556,
        0.13366802,
        0.01932214,
        0.09104561,
        -0.064578414,
        0.4198957,
        -0.047274455,
        0.002733558,
        -0.01195226,
        0.012568232,
        0.070971005,
        -0.10084235,
        -0.04704455,
        0.016822495,
        0.021196425,
        0.010339204,
        -0.023591692,
        -0.0058193714,
        0.02593006,
        0.012279002,
        -0.051120713,
        -0.030370984,
        -0.097752444,
        -0.003641293,
        -0.019987265,
        0.017142707,
        -0.07959971,
        -0.0004898602,
        0.026605425,
        -0.04501932,
        -0.0042946436,
        0.048095956,
        0.03487082,
        -0.12147835,
        0.08539147,
        -0.0045770877,
        -0.0762686,
        -0.03809889,
        -0.04100646,
        0.049712956,
        -0.022192903,
        -0.029919062,
        0.123707004,
        -0.04168466,
        0.052348495,
        -0.07038362,
        0.020794386,
        0.012369724,
        -0.005432135,
        0.037194446,
        -0.06547619,
        -0.15368995,
        -0.049824335,
        0.030105023,
        0.042600594,
        0.010647169,
        -0.004894343,
        0.08989774,
        0.007369944,
        -0.021035412,
        0.075825214,
        -0.055888608,
        -0.035736855,
        0.041227765,
        -0.058219783,
        0.09130208,
        0.06487335,
        -0.040270884,
        -0.07444561,
        -0.013101277,
        -0.050640997,
        0.023403417,
        -0.08690528,
        0.019100914,
        -0.015100144,
        0.015871739,
        0.024199637,
        -0.061823383,
        0.061797284,
        -0.061829243,
        -0.073855855,
        -0.044262365,
        0.00365496,
        -0.019091872,
        0.054088272,
        -0.0014424775,
        -0.007214397,
        -0.61603856,
        -0.05292063,
        0.010585981,
        -0.07949842,
        -0.17640346,
        0.010506919,
        0.015392932,
        0.09045461,
        0.086909644,
        0.002659703,
        0.09107156,
        0.0033285841,
        0.03860786,
        -0.0014358102,
        0.018618843,
        0.05900405,
        0.022856187
      ]
    },
    {
      "id": "worldsfair-speaker-275",
      "conference": "worldsfair",
      "name": "Stephanie Jarmak",
      "role": "",
      "company": "Sourcegraph",
      "twitter": "",
      "talks": [
        {
          "title": "The Death of Developer Advocates",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        -0.018131789,
        0.018932074,
        0.02695211,
        -0.06315019,
        -0.0058518276,
        -0.0075283214,
        -0.06062454,
        -0.036048774,
        0.07951842,
        -0.17394586,
        -0.055238195,
        -0.06056822,
        0.0015885048,
        0.009450679,
        -0.076719664,
        0.045160178,
        0.08845848,
        -0.108963825,
        -0.0035080092,
        -0.084935226,
        0.018197345,
        -0.05974601,
        0.04147272,
        0.031183403,
        -0.044078525,
        0.1294463,
        0.042501613,
        0.01145293,
        -0.12632732,
        0.48970214,
        -0.059593655,
        0.027807605,
        -0.0013374769,
        -0.056247957,
        0.063900635,
        -0.041775756,
        -0.010257475,
        0.023753772,
        0.04509947,
        -0.008459436,
        -0.024778526,
        0.0513309,
        -0.07095141,
        0.03414746,
        -0.011594033,
        0.01757171,
        -0.11726811,
        -0.03909312,
        0.004254736,
        0.043457594,
        -0.07663818,
        0.0168747,
        0.002482908,
        -0.0498649,
        0.035184164,
        -0.004749735,
        -0.028820265,
        -0.044693083,
        0.105159715,
        -0.083450146,
        -0.12138223,
        0.009615597,
        -0.036111124,
        0.017397447,
        0.04515853,
        -0.08940403,
        0.1152839,
        -0.015804546,
        -0.0077606714,
        -0.047838625,
        0.025900492,
        -0.03516765,
        -0.06124531,
        0.0666182,
        -0.051158655,
        -0.07604208,
        0.022996321,
        0.06977327,
        -0.027603673,
        0.046142388,
        0.022738824,
        0.005397511,
        -0.1083627,
        -0.050001033,
        0.008668789,
        -0.02232854,
        -0.010317542,
        0.059704937,
        -0.049890902,
        -0.00938827,
        0.0007352101,
        -0.018673778,
        0.005221415,
        0.0344034,
        -0.027841248,
        0.0037407014,
        -0.07260726,
        0.08949003,
        0.06583537,
        0.03723233,
        -0.012447002,
        -0.04276039,
        0.00033547686,
        -0.05357572,
        -0.009974259,
        -0.05853962,
        -0.019068629,
        0.01215899,
        0.021407483,
        0.036264606,
        0.022916952,
        -0.59297585,
        -0.019557942,
        0.032024067,
        0.032730788,
        -0.1575098,
        -0.05459563,
        -0.044322208,
        0.043001138,
        0.15289108,
        -0.058004506,
        0.045559194,
        -0.02680924,
        0.03465932,
        0.013310241,
        0.0749828,
        0.091026306,
        0.023359025
      ]
    },
    {
      "id": "worldsfair-speaker-276",
      "conference": "worldsfair",
      "name": "Rashi Agrawal",
      "role": "Head of Agentic AI",
      "company": "Hinge Health",
      "twitter": "",
      "talks": [
        {
          "title": "Guardrails First: Engineering Member-Facing Health AI",
          "description": "Everywhere else in the company, an AI pilot can reach production in weeks. For our member-facing clinical assistant, it can't, and that single constraint redesigned our entire architecture. This is a field report on building conversational AI in a regulated digital health setting, where \"move fast and break things\" isn't a culture choice. It's a liability. We'll get concrete about what changes when every output has to be clinically safe, auditable, and compliant: PHI is protected by architecture, not policy. Production and non-production are hard-isolated, dashboards are sanitized, and engineers outside the US never touch protected health information. Must-not-fail behavior never lives in a prompt. Emergency escalation and intent routing run as deterministic rules at the top of every conversation turn, before the model is consulted. If you can't afford to get something wrong, you don't leave it to a probabilistic system. Clinical safety is a continuous eval layer. ~30 LLM-as-judge evaluators score clinical accuracy, clinical safety, escalation routing, and recommendation relevance, continuously, not once. Every output is auditable. Each turn, tool call, and reasoning step is traced so outputs can be reviewed and meet regulated reporting obligations. The throughline: in regulated healthcare, compliance constraints aren't a tax you pay around the architecture. They become the architecture. We'll talk about why guardrails-first is the only way to ship member-facing health AI, and why \"painfully slow\" is sometimes exactly right. (This is non-diagnostic, member-facing AI. The talk is about engineering discipline under regulation, not medical claims.) Key takeaways - In regulated health AI, \"move fast\" is the wrong default. Design for deliberate, careful launches. - Must-not-fail behaviors belong in deterministic rules at the top of every turn, never in the prompt. - Protect PHI through architecture: isolate prod from non-prod, sanitize dashboards, restrict access by role and geography. - Make every output auditable. Trace each turn, tool call, and reasoning step so safety is reviewable, not assumed. - Treat clinical safety as a continuous LLM-as-judge layer, not a one-time gate.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.0127062965,
        0.0844027,
        0.044275172,
        0.0037152297,
        -0.008218943,
        0.009292405,
        -0.090162136,
        -0.05013316,
        0.044674914,
        -0.22839688,
        -0.022832192,
        -0.04537488,
        0.045531657,
        -0.070417486,
        -0.048849992,
        0.025854517,
        0.09626608,
        -0.07151761,
        -0.027016839,
        -0.043175165,
        -0.017784195,
        0.00791609,
        -0.008577179,
        0.11582357,
        -0.05468709,
        0.027957506,
        0.0005510422,
        0.052715085,
        -0.12925553,
        0.451901,
        -0.12131221,
        0.009790716,
        0.025651628,
        -0.013883005,
        0.054626014,
        -0.05213633,
        -0.0009917845,
        -0.07518879,
        0.015041456,
        0.06576564,
        0.005453594,
        0.03403009,
        -0.050199285,
        0.02427731,
        -0.050673466,
        -0.046208672,
        -0.053494107,
        0.027724596,
        -0.027890332,
        -0.0064802007,
        -0.069337,
        -0.015442967,
        0.042215552,
        -0.04709175,
        0.009138999,
        0.05302794,
        -0.008618597,
        -0.08798608,
        0.08086708,
        -0.017644173,
        -0.053589743,
        0.004114986,
        0.037306715,
        -0.03395006,
        0.05701542,
        -0.092546225,
        0.06582823,
        -0.031592283,
        -0.041353777,
        0.036176205,
        0.007428111,
        -0.0023252668,
        -0.078687675,
        0.04437218,
        -0.09026446,
        -0.13731933,
        -0.04964477,
        0.0761864,
        -0.009928463,
        0.014894931,
        0.0056702937,
        0.026156748,
        -0.062143266,
        -0.031118868,
        0.06356794,
        -0.04229009,
        -0.032104958,
        0.06089677,
        -0.009772497,
        0.033355158,
        0.028933242,
        -0.05538051,
        -0.056167196,
        -0.004940829,
        -0.06847827,
        0.030146899,
        -0.05677619,
        -0.0027368704,
        0.0015147334,
        0.072303005,
        0.012304604,
        -0.07581241,
        0.014764803,
        0.0013922071,
        -0.015335799,
        -0.0052795387,
        -0.0063625346,
        -0.02060556,
        0.052716184,
        -0.005013286,
        -0.059099108,
        -0.6210097,
        0.0106050605,
        0.014306212,
        -0.018099226,
        -0.16324732,
        0.09015933,
        -0.016109342,
        0.00045323628,
        0.12788847,
        -0.06904452,
        0.09323477,
        -0.08665968,
        -0.017620696,
        0.002012133,
        -0.009828413,
        0.05486517,
        -0.008048146
      ]
    },
    {
      "id": "worldsfair-speaker-277",
      "conference": "worldsfair",
      "name": "Qianru Lao",
      "role": "Member of Technical Staff, Inference",
      "company": "OpenAI",
      "twitter": "",
      "talks": [
        {
          "title": "Routing LLM Inference in Production: From Engine Signals to Policy",
          "description": "Production LLM apps need more than a fast model: they need an inference routing layer that can choose where each request should run as engines, capacity, latency, and geography cost change. This talk shares a generalized Inference Load Balancer (ILB) proxy/controller architecture. A low-latency proxy applies routing weights and request-path signals, while a controller computes source-cluster-to-engine weights from demand, capacity/performance profiles, replica state, and geography cost. We will cover the practical debugging patterns AI engineers need: reading engine signals, explaining why a request went to one backend instead of another, handling retries and load shedding, and keeping routing behavior observable without exposing OpenAI-specific internals or non-public metrics.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.05639356,
        0.057414066,
        0.0045961905,
        0.047342766,
        0.042995982,
        -0.080455855,
        -0.079656005,
        0.03272402,
        0.03380132,
        -0.22338107,
        -0.0039590215,
        -0.056489497,
        -0.031775728,
        -0.04664849,
        -0.071713746,
        0.036290854,
        0.1078445,
        0.0017726691,
        -0.07179114,
        -0.069113895,
        -0.049321443,
        -0.07851644,
        0.014368935,
        0.031028597,
        -0.025490917,
        0.08406305,
        0.059232272,
        -0.0016558123,
        -0.13364702,
        0.39090052,
        -0.18757969,
        0.0067443647,
        -0.025624562,
        -0.03282558,
        0.0920599,
        -0.05011179,
        0.021651732,
        -0.05521859,
        0.032417938,
        0.01884318,
        0.017856937,
        0.05101704,
        -0.06694366,
        0.038400117,
        -0.07254302,
        -0.005183372,
        -0.0662801,
        0.0108584855,
        -0.0059463056,
        0.040721178,
        -0.01653319,
        -0.032772303,
        0.012980317,
        -0.009192937,
        0.049738374,
        0.04047788,
        -0.05700696,
        -0.06824777,
        0.11211507,
        -0.024001418,
        -0.11198058,
        0.061063673,
        0.037409045,
        0.0020237088,
        0.05106496,
        -0.083233066,
        0.033986095,
        -0.07437792,
        -0.024086704,
        0.022955773,
        0.048442364,
        -0.0012174121,
        -0.03903346,
        0.007820954,
        -0.06599249,
        -0.1261898,
        0.01636417,
        0.050714817,
        0.005493842,
        -0.027448367,
        0.0523974,
        0.13499972,
        -0.038346056,
        -0.024916977,
        0.039612953,
        -0.04163839,
        -0.0844804,
        0.041045602,
        -0.06890058,
        0.09260354,
        0.025171068,
        -0.060085878,
        -0.057874467,
        -0.0010940965,
        -0.097064525,
        0.009381918,
        -0.054827113,
        0.058484193,
        -0.02680199,
        0.09663521,
        0.06438408,
        0.007994541,
        0.028557999,
        0.021519965,
        -0.07621898,
        -0.01616579,
        0.0022272412,
        0.0044306153,
        0.074840605,
        -0.00041726252,
        -0.084542654,
        -0.614122,
        0.042391136,
        0.029000178,
        -0.03670064,
        -0.13033713,
        0.07734147,
        -0.06217338,
        -0.04293193,
        0.03333688,
        -0.085158594,
        0.04370745,
        -0.053472854,
        -0.057929773,
        0.033165976,
        0.024504878,
        0.008257641,
        0.03257255
      ]
    },
    {
      "id": "worldsfair-speaker-278",
      "conference": "worldsfair",
      "name": "Lu Zhang",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Routing LLM Inference in Production: From Engine Signals to Policy",
          "description": "Production LLM apps need more than a fast model: they need an inference routing layer that can choose where each request should run as engines, capacity, latency, and geography cost change. This talk shares a generalized Inference Load Balancer (ILB) proxy/controller architecture. A low-latency proxy applies routing weights and request-path signals, while a controller computes source-cluster-to-engine weights from demand, capacity/performance profiles, replica state, and geography cost. We will cover the practical debugging patterns AI engineers need: reading engine signals, explaining why a request went to one backend instead of another, handling retries and load shedding, and keeping routing behavior observable without exposing OpenAI-specific internals or non-public metrics.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.0398189,
        0.067600116,
        0.025423782,
        0.038314678,
        0.07624742,
        -0.0788272,
        -0.059011765,
        0.045281254,
        0.06273367,
        -0.251395,
        0.0005340328,
        -0.0384185,
        0.022627348,
        -0.03813608,
        -0.07976267,
        0.03743892,
        0.09834999,
        0.0077987015,
        -0.10997239,
        -0.060198992,
        -0.064116955,
        -0.08277973,
        -0.017861055,
        0.030581424,
        -0.036818758,
        0.10736496,
        0.02962128,
        0.008360901,
        -0.118924655,
        0.37805486,
        -0.09260995,
        -0.016261663,
        -0.034227036,
        -0.03281521,
        0.004697072,
        -0.068980314,
        -0.000014672617,
        -0.028642192,
        0.06786568,
        0.046406038,
        0.05610949,
        0.027415486,
        -0.05913776,
        0.013579468,
        -0.07016635,
        -0.037311926,
        -0.10004021,
        0.0016911032,
        -0.008185559,
        0.061788894,
        -0.03231373,
        -0.026729096,
        0.037343662,
        -0.0002523827,
        -0.025596784,
        0.03115253,
        0.0029025222,
        -0.08787303,
        0.10042366,
        -0.015569356,
        -0.13728718,
        0.031102695,
        -0.0037628093,
        -0.018738478,
        0.057777066,
        -0.051552497,
        -0.0038896136,
        -0.08894372,
        -0.06223491,
        0.009002283,
        0.042545304,
        -0.028468125,
        -0.041884027,
        0.0032710277,
        -0.06868192,
        -0.15717946,
        0.012726124,
        0.031262126,
        -0.010240486,
        0.028915877,
        0.0053693694,
        0.079630606,
        -0.020251304,
        -0.016753905,
        0.008118891,
        0.025434474,
        -0.08705413,
        0.07051928,
        -0.081518546,
        0.1376451,
        0.029425347,
        0.0065905657,
        -0.021071011,
        -0.02026482,
        -0.036555924,
        0.03946976,
        -0.047577754,
        0.06897566,
        -0.022176653,
        0.08238592,
        0.046861507,
        0.052243963,
        0.009735386,
        0.014970149,
        -0.01883949,
        -0.073644735,
        0.0073947646,
        0.0046878927,
        0.004823609,
        -0.010006176,
        -0.0783115,
        -0.63778865,
        -0.007231102,
        0.04248402,
        -0.034816314,
        -0.13813938,
        0.08840764,
        -0.016318146,
        -0.014001243,
        0.035558056,
        -0.05784232,
        0.07547108,
        -0.05059742,
        -0.057420697,
        0.022882683,
        0.02020757,
        0.027573556,
        0.044122398
      ]
    },
    {
      "id": "worldsfair-speaker-279",
      "conference": "worldsfair",
      "name": "Dmitry Buykin",
      "role": "",
      "company": "Maersk",
      "twitter": "",
      "talks": [
        {
          "title": "Tribal Dungeons of Global Shipping CX: AI Agents at 100K Cases a Day",
          "description": "Most \"AI agents in production\" talks skip the part where you have to drag tribal knowledge out of 100+ country SME teams and turn it into something an agent can execute safely. This is that part. How Maersk ships AI agents handling 100K customer cases a day across global logistics, and why extracting and aligning the tribal knowledge was 10x harder than the agent itself. - Why SOPs-as-code (versioned markdown, per-country) beats prompt engineering at this scale - The SME alignment loop: how corrections become SOP changes without breaking 99 other countries - Guardrails that matter in production: write-gating, loop breakers, classifier vs. SOP-body routing layers - Where agents under-deliver against the demo, and how we measured it honestly - Org/process patterns for the Applied AI / Forward Deployed Engineer stack across 100+ countries",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.026503155,
        0.007492844,
        0.021240015,
        0.0257772,
        -0.04359879,
        -0.0046549197,
        -0.059390977,
        0.020410223,
        0.040969986,
        -0.20458321,
        -0.049854167,
        -0.00437194,
        0.054141514,
        -0.019479813,
        -0.06938616,
        -0.010767925,
        0.13423827,
        -0.023377927,
        -0.046159327,
        -0.095654726,
        -0.033111375,
        -0.10441151,
        -0.023087298,
        0.0035679364,
        -0.024739683,
        0.087167874,
        -0.018872887,
        0.0969412,
        -0.08032032,
        0.48841763,
        -0.1411098,
        0.03913234,
        0.026638443,
        -0.037333615,
        0.022950087,
        -0.07955933,
        -0.053961612,
        -0.06876671,
        0.013449213,
        0.06579865,
        0.033474326,
        0.09036466,
        -0.025705058,
        0.04113838,
        -0.03289619,
        -0.01564727,
        -0.028191501,
        0.009102768,
        0.016135788,
        -0.1086234,
        -0.04317555,
        -0.025187762,
        0.061013643,
        -0.012102132,
        0.053126246,
        0.038856134,
        0.06494704,
        -0.024796717,
        0.11514717,
        -0.08708557,
        -0.072313756,
        0.0015810439,
        -0.010406617,
        -0.0132418405,
        0.0074234935,
        -0.07310051,
        0.0062438543,
        -0.048602507,
        0.016265973,
        -0.011604928,
        -0.01724999,
        0.020294348,
        -0.06961561,
        0.07421968,
        -0.041757498,
        -0.12270822,
        -0.03456154,
        0.10810371,
        -0.03236876,
        -0.051331487,
        -0.053098805,
        0.095702894,
        -0.07334269,
        -0.068152726,
        0.053173132,
        -0.019457173,
        -0.062536925,
        0.018778821,
        -0.04440917,
        0.055833492,
        0.056832116,
        0.0029537256,
        -0.06878337,
        0.0023529343,
        -0.07039297,
        0.04431166,
        0.002295177,
        0.020893035,
        -0.027383147,
        0.028708698,
        -0.03636877,
        -0.058383,
        0.027942324,
        0.025208775,
        0.047978055,
        -0.0027005891,
        0.0031570601,
        -0.039253782,
        0.087195605,
        -0.07103316,
        0.011223055,
        -0.59251046,
        -0.05557409,
        0.031890485,
        0.00027278234,
        -0.1437283,
        0.03292577,
        -0.007397104,
        0.02240081,
        0.07717422,
        0.0014381387,
        0.05564448,
        -0.034246292,
        -0.02867959,
        -0.0067676,
        0.049183168,
        0.059539113,
        0.029335322
      ]
    },
    {
      "id": "worldsfair-speaker-280",
      "conference": "worldsfair",
      "name": "Tisha Chawla",
      "role": "Software Engineer",
      "company": "Microsoft",
      "twitter": "",
      "talks": [
        {
          "title": "FinOps for AI Agents: Who Spent All the Tokens?",
          "description": "When an autonomous agent finishes a task successfully but costs ten times more than it did the previous day, traditional application monitoring fails. A recursive tool loop that retries silently, an oversized context window that quietly expands, or an unflagged model upgrade can burn through an entire budget long before a human notices. The execution appears successful on functional dashboards, meaning the only clear signal of failure is the cloud invoice at the end of the month. As AI systems move into production, tokens have become a primary operational resource alongside CPU, memory, and storage, yet few teams manage them with equivalent systems rigor. Most architectures lack the granular visibility required to attribute token spend to specific users, agents, or workflows, and they lack mechanisms to terminate a runaway loop before it triggers a financial incident. This session treats token consumption as a first class systems problem, demonstrating how to make it observable, attributable, and enforceable across complex agent workflows. The presentation covers practical engineering patterns for instrumenting token usage at every model call and tool invocation, attributing costs down to specific users or business operations, surfacing expensive execution paths, and enforcing runtime budgets, quotas, and circuit breakers to halt runaway behavior in real time. Attendees will leave with a practical framework for governing agent spend deliberately, transforming tokens into a managed operational resource rather than a surprise line item on the cloud bill.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.08889107,
        0.055898093,
        0.030911881,
        0.02258273,
        0.008971354,
        -0.033194277,
        -0.0791644,
        0.00011644613,
        0.054441147,
        -0.278798,
        -0.02701044,
        -0.029084058,
        0.0001477449,
        -0.08772238,
        -0.05943827,
        0.005510897,
        0.0786846,
        -0.067551605,
        -0.033419717,
        -0.046657763,
        -0.032479,
        -0.053576272,
        -0.040392235,
        0.057197858,
        -0.04334122,
        0.110814475,
        0.024807,
        0.012673665,
        -0.110460065,
        0.48458222,
        -0.15521298,
        0.02154424,
        0.029735906,
        -0.014894947,
        0.0035751555,
        -0.030903364,
        -0.029471382,
        0.02970911,
        0.027638888,
        0.022195738,
        -0.0009774318,
        0.035398595,
        -0.030514816,
        0.047809366,
        -0.031222973,
        -0.0804935,
        -0.09261019,
        0.03144267,
        -0.054276317,
        0.033275645,
        -0.03223395,
        -0.024539728,
        0.04558643,
        -0.008132783,
        0.08172854,
        0.07899907,
        -0.020255392,
        -0.075030364,
        0.093455344,
        -0.06397125,
        -0.06802002,
        0.0007708248,
        -0.02901192,
        -0.011294326,
        0.025317028,
        -0.07053438,
        0.06692934,
        -0.064585924,
        -0.046558384,
        -0.052072093,
        0.052960202,
        0.0476941,
        -0.020260265,
        0.06439684,
        -0.008685508,
        -0.15476239,
        -0.03410451,
        0.09610003,
        -0.0017656339,
        0.03400956,
        0.014238017,
        0.03711876,
        -0.11322723,
        -0.03725185,
        0.035374064,
        0.009784869,
        -0.06376368,
        0.013322518,
        -0.043652713,
        0.037750896,
        -0.007861043,
        -0.008227532,
        -0.054581743,
        0.038575757,
        -0.11500265,
        -0.0011224928,
        -0.04512599,
        0.05878813,
        -0.028102672,
        -0.002223297,
        0.00089644897,
        -0.05485387,
        0.005739797,
        -0.011537063,
        -0.048542876,
        -0.034104034,
        0.011876349,
        0.038693137,
        0.04948007,
        0.027606554,
        -0.040978216,
        -0.569035,
        0.014374608,
        0.00045346897,
        -0.0874281,
        -0.15768246,
        0.0528608,
        -0.026571052,
        0.02709882,
        0.098777235,
        -0.014311944,
        0.051905483,
        -0.060688637,
        -0.034459066,
        -0.012436784,
        0.0029527901,
        0.015208302,
        0.016402684
      ]
    },
    {
      "id": "worldsfair-speaker-281",
      "conference": "worldsfair",
      "name": "Susheem Koul",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "FinOps for AI Agents: Who Spent All the Tokens?",
          "description": "When an autonomous agent finishes a task successfully but costs ten times more than it did the previous day, traditional application monitoring fails. A recursive tool loop that retries silently, an oversized context window that quietly expands, or an unflagged model upgrade can burn through an entire budget long before a human notices. The execution appears successful on functional dashboards, meaning the only clear signal of failure is the cloud invoice at the end of the month. As AI systems move into production, tokens have become a primary operational resource alongside CPU, memory, and storage, yet few teams manage them with equivalent systems rigor. Most architectures lack the granular visibility required to attribute token spend to specific users, agents, or workflows, and they lack mechanisms to terminate a runaway loop before it triggers a financial incident. This session treats token consumption as a first class systems problem, demonstrating how to make it observable, attributable, and enforceable across complex agent workflows. The presentation covers practical engineering patterns for instrumenting token usage at every model call and tool invocation, attributing costs down to specific users or business operations, surfacing expensive execution paths, and enforcing runtime budgets, quotas, and circuit breakers to halt runaway behavior in real time. Attendees will leave with a practical framework for governing agent spend deliberately, transforming tokens into a managed operational resource rather than a surprise line item on the cloud bill.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.08729041,
        0.07397482,
        0.03691052,
        -0.017758125,
        0.036916807,
        -0.027862636,
        -0.014258256,
        -0.020970518,
        0.035138436,
        -0.2492809,
        -0.0024675517,
        -0.040752478,
        -0.017520202,
        -0.060254212,
        -0.05119141,
        -0.05278838,
        0.0979379,
        -0.11858102,
        -0.056297984,
        -0.05082371,
        -0.028803179,
        -0.08101914,
        0.023368353,
        0.06163558,
        -0.03988833,
        0.06401049,
        -0.043215975,
        0.0634599,
        -0.09437606,
        0.46077588,
        -0.06951925,
        -0.015371532,
        0.031198384,
        -0.038779594,
        -0.06666181,
        -0.08765281,
        -0.024891825,
        0.010190519,
        0.000018176313,
        0.086843126,
        -0.013496157,
        0.056144137,
        0.0028988568,
        0.02778858,
        -0.0663301,
        -0.12348759,
        -0.049990565,
        -0.014257878,
        -0.007652163,
        -0.012958546,
        -0.03510645,
        -0.03754752,
        0.0009686411,
        -0.01794786,
        0.05316754,
        0.017743533,
        -0.005697176,
        -0.10796439,
        0.08893875,
        -0.04008627,
        -0.095664,
        0.039663035,
        -0.10121222,
        0.020130249,
        0.0030408106,
        -0.040084574,
        0.08878005,
        -0.055495866,
        -0.0050551808,
        -0.023596909,
        -0.010772759,
        -0.014358514,
        -0.044085898,
        0.082940154,
        -0.05140603,
        -0.15124746,
        -0.030382155,
        0.08427821,
        -0.07176982,
        0.037681624,
        -0.017268814,
        0.08461275,
        -0.094636604,
        -0.043073338,
        0.0025984212,
        -0.011874051,
        -0.035759687,
        0.00968375,
        0.0022502008,
        0.05291355,
        0.014577892,
        0.015920665,
        -0.066318944,
        0.040841382,
        -0.06950893,
        0.021106495,
        -0.07551321,
        0.05246122,
        -0.039747063,
        -0.0038501793,
        0.007208481,
        -0.061665595,
        -0.005973636,
        -0.04919721,
        0.0022738627,
        -0.052050874,
        0.026294952,
        0.015557878,
        0.011557381,
        0.004599562,
        0.00032876994,
        -0.5929282,
        -0.006998626,
        0.038149398,
        -0.12176376,
        -0.15705805,
        0.068963416,
        -0.004973316,
        0.071563154,
        0.059730798,
        -0.018087527,
        0.032161742,
        -0.035256702,
        -0.027778085,
        -0.041714747,
        0.03418384,
        0.031405583,
        -0.022490392
      ]
    },
    {
      "id": "worldsfair-speaker-282",
      "conference": "worldsfair",
      "name": "Michael Patterson",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Lethal Trifecta Is Already on Your Developers' Laptops",
          "description": "The lethal trifecta: an AI agent with access to private data, exposure to untrusted content, and the ability to communicate externally. Combine all three and an attacker can trick your agent into exfiltrating anything it can see and there is no prompt-level fix.. Most enterprises have already deployed this pattern at scale: Claude Code, Cursor, and Copilot on developer laptops with local credentials, MCPs reaching into internal systems, and open egress. I'll speak to my own personal agent stack as a textbook example, then trace the same shape across enterprise deployments I see at Coder. The back half is four architectural moves that defuse it: governed compute, centralized credentials, default-deny egress, identity-bound audit. Walk out with a mental model and a checklist you can run against your own deployment the next morning.",
          "day": "Day 4 — Session Day 3",
          "time": "11:10am-11:30am",
          "type": "session"
        }
      ],
      "embedding": [
        0.041777417,
        0.032667853,
        0.05355214,
        -0.022575011,
        0.053863723,
        -0.013266512,
        -0.0063148467,
        -0.004280359,
        0.015014687,
        -0.22136782,
        -0.026991216,
        -0.017516708,
        -0.018495388,
        -0.050557777,
        -0.037381675,
        -0.04265803,
        0.098283604,
        -0.040949594,
        -0.07053323,
        -0.02331316,
        -0.05652028,
        -0.13008599,
        0.0054945387,
        0.008181291,
        -0.050802056,
        0.15353577,
        -0.04590677,
        0.106632076,
        -0.053100877,
        0.42194802,
        -0.032304913,
        0.0887883,
        0.029964354,
        -0.029454026,
        0.0013589931,
        -0.041932076,
        -0.034178033,
        0.041309953,
        -0.01859302,
        0.050927434,
        -0.020890312,
        0.020254813,
        -0.064755216,
        0.040591713,
        -0.02132738,
        0.0042680977,
        -0.12529305,
        -0.033835012,
        0.07166599,
        -0.07197769,
        -0.029045995,
        -0.020769436,
        0.11183071,
        -0.037166893,
        0.031100875,
        0.08777209,
        0.00029599978,
        -0.032148954,
        0.11574639,
        -0.10089307,
        -0.022455836,
        -0.015693143,
        -0.011138045,
        0.023288898,
        -0.04532778,
        -0.04748537,
        0.10572402,
        -0.029015468,
        0.04801691,
        -0.04187056,
        0.006957679,
        0.0327972,
        -0.01303751,
        0.07971297,
        -0.07640772,
        -0.11336976,
        -0.11651027,
        0.026706768,
        -0.0758573,
        0.09543684,
        -0.0073982636,
        0.06979926,
        -0.023133466,
        -0.08111322,
        -0.0031177928,
        0.04799632,
        -0.08000334,
        0.024285,
        -0.0134776,
        0.026488012,
        0.0064549902,
        0.02039138,
        -0.013745224,
        0.058317464,
        -0.05759785,
        0.036894526,
        -0.030376006,
        0.05452108,
        0.017858386,
        -0.02172992,
        -0.05655577,
        -0.062054697,
        0.015508181,
        -0.0393263,
        0.052478362,
        -0.03339758,
        0.030903395,
        -0.041122764,
        0.008579664,
        0.00042793635,
        0.009548933,
        -0.62764925,
        -0.008526659,
        -0.02669866,
        -0.04777307,
        -0.1546584,
        0.050962552,
        -0.012723189,
        0.039299786,
        0.07962199,
        -0.044992883,
        0.046037115,
        -0.00739848,
        -0.009800005,
        0.013680195,
        0.032325745,
        0.097226016,
        -0.026418583
      ]
    },
    {
      "id": "worldsfair-speaker-283",
      "conference": "worldsfair",
      "name": "Maxime Rivest",
      "role": "",
      "company": "DSPy",
      "twitter": "",
      "talks": [
        {
          "title": "The Unreasonable Effectiveness of Separating the Task from the Model",
          "description": "By declaring your task’s inputs and outputs without initially considering model capability, you create the space needed to figure out the model execution later. DSPy’s entire promise is that you should evaluate and execute your AI engineering at a level higher than a specific prompt template or a particular provider’s API shape: the Signature. However, models have evolved significantly over the last few years. How can the same input and output specifications still work in a world now filled with tools, RLMs, and Skills? By defining your task strictly through its inputs and outputs, the underlying implementation becomes completely flexible. You can experiment with different models, settings, weights, templating strategies, and output formats, all without touching your actual AI workflow. Consequently, you can leverage components built by others and focus entirely on your core AI task. In this talk we will present how dspy 3.5 makes it easier much easier. DSPy has its roots in prompt optimization, where we build efficient ways to conduct search and learning beneath the signature. In this talk we will give a preview of DSPy 4.0 where we use the fact that models have now passed a tipping point for two critical concepts we have always needed. First, we no longer need to limit the search space to a single instruction block per LLM call; models can now reliably write the code underneath a signature themselves—so they should. Second, traditional prompt optimization has always required a scalar metric, which is notoriously one of the hardest parts to get right. What if a DSPy program could learn directly from your interactions with users? Ultimately, all you care about is that the function you call respects the inputs and outputs of your signature. You can let the models figure out the rest.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0013257429,
        0.023231536,
        0.03666579,
        0.0038435075,
        0.043565184,
        -0.025922686,
        -0.048831347,
        -0.0073119923,
        0.022376452,
        -0.21545106,
        -0.03338607,
        -0.0197477,
        -0.010925467,
        -0.036234442,
        -0.06865784,
        -0.0017736115,
        0.15366578,
        -0.08742135,
        -0.028668894,
        -0.069347374,
        -0.048719853,
        -0.123929024,
        -0.06296946,
        0.04327289,
        -0.042791557,
        0.071537934,
        -0.010517487,
        0.013900405,
        -0.11666281,
        0.4728747,
        -0.12677622,
        0.030941421,
        -0.0120141925,
        0.019986942,
        0.023833223,
        -0.08948494,
        -0.027147826,
        0.04332679,
        0.0057281586,
        0.047867,
        -0.04279235,
        0.040898204,
        -0.081692666,
        0.05852201,
        0.0052606207,
        0.0102464575,
        -0.10428177,
        -0.010179683,
        -0.011971141,
        -0.04285484,
        -0.04793502,
        0.041033156,
        0.008880352,
        -0.05961257,
        0.050272733,
        0.10427968,
        0.019959588,
        -0.065578215,
        0.13454492,
        -0.07944084,
        -0.13056478,
        -0.036324102,
        0.0066727116,
        0.06448734,
        -0.00224826,
        -0.06520743,
        0.05780381,
        -0.029730322,
        -0.023531122,
        -0.05551486,
        0.023521895,
        -0.028113754,
        -0.086118035,
        0.022762015,
        -0.11124272,
        -0.031581778,
        -0.023547059,
        0.057238963,
        0.009213843,
        0.03741552,
        -0.011493466,
        0.046657797,
        -0.04573556,
        -0.081160754,
        0.08224272,
        0.014802835,
        -0.00015888226,
        -0.006910995,
        -0.022266341,
        0.073443085,
        0.0048983116,
        -0.024838192,
        -0.018016461,
        -0.0050403494,
        -0.049804796,
        -0.076527946,
        -0.004257175,
        0.07037285,
        -0.0029114843,
        0.013750144,
        0.039694536,
        -0.025094243,
        0.12551136,
        0.0038304587,
        -0.007779296,
        -0.034677662,
        -0.008660131,
        -0.03096827,
        0.03154541,
        -0.03898436,
        -0.059277035,
        -0.6024831,
        0.004731663,
        -0.02366848,
        -0.03292224,
        -0.15044373,
        0.018275192,
        -0.022353308,
        -0.006785302,
        0.036727257,
        -0.004025329,
        0.060725853,
        -0.006844871,
        -0.025433237,
        0.034882076,
        0.035597514,
        0.03881588,
        0.020505236
      ]
    },
    {
      "id": "worldsfair-speaker-284",
      "conference": "worldsfair",
      "name": "Isaac Miller",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Unreasonable Effectiveness of Separating the Task from the Model",
          "description": "By declaring your task’s inputs and outputs without initially considering model capability, you create the space needed to figure out the model execution later. DSPy’s entire promise is that you should evaluate and execute your AI engineering at a level higher than a specific prompt template or a particular provider’s API shape: the Signature. However, models have evolved significantly over the last few years. How can the same input and output specifications still work in a world now filled with tools, RLMs, and Skills? By defining your task strictly through its inputs and outputs, the underlying implementation becomes completely flexible. You can experiment with different models, settings, weights, templating strategies, and output formats, all without touching your actual AI workflow. Consequently, you can leverage components built by others and focus entirely on your core AI task. In this talk we will present how dspy 3.5 makes it easier much easier. DSPy has its roots in prompt optimization, where we build efficient ways to conduct search and learning beneath the signature. In this talk we will give a preview of DSPy 4.0 where we use the fact that models have now passed a tipping point for two critical concepts we have always needed. First, we no longer need to limit the search space to a single instruction block per LLM call; models can now reliably write the code underneath a signature themselves—so they should. Second, traditional prompt optimization has always required a scalar metric, which is notoriously one of the hardest parts to get right. What if a DSPy program could learn directly from your interactions with users? Ultimately, all you care about is that the function you call respects the inputs and outputs of your signature. You can let the models figure out the rest.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.048995856,
        0.029243257,
        0.058741484,
        0.014276241,
        0.042758875,
        -0.094839945,
        0.004258346,
        -0.009242128,
        0.0787927,
        -0.2569685,
        -0.020212367,
        -0.034210827,
        -0.035675235,
        -0.06590323,
        -0.062450297,
        -0.0028911664,
        0.14393537,
        -0.09487449,
        -0.03482138,
        -0.06900501,
        -0.027974458,
        -0.15055762,
        -0.016068814,
        0.060693353,
        -0.039305273,
        0.07530736,
        -0.043228243,
        0.06830891,
        -0.097195394,
        0.41566858,
        -0.060095575,
        0.050448216,
        0.0031846608,
        0.02325599,
        -0.0061253225,
        -0.04965174,
        -0.02521891,
        0.05310722,
        -0.0076047247,
        0.028471544,
        -0.005560754,
        0.026015276,
        -0.019510958,
        0.08155661,
        -0.018436367,
        -0.06730441,
        -0.04785121,
        0.018226808,
        -0.028789094,
        -0.020803189,
        -0.054608796,
        0.04461351,
        0.034218475,
        -0.07734311,
        0.038393244,
        0.07520692,
        -0.043119382,
        -0.06657064,
        0.17537183,
        -0.031140495,
        -0.058331832,
        -0.019609738,
        -0.017207293,
        0.05899106,
        0.057387106,
        -0.048383087,
        0.079780065,
        -0.024265558,
        -0.0028227814,
        -0.07474232,
        0.025089024,
        -0.027377004,
        -0.06372086,
        0.086011976,
        -0.11619477,
        -0.11104771,
        -0.0028624188,
        0.012931081,
        -0.013635372,
        0.03161115,
        0.010692018,
        0.07798368,
        -0.033343762,
        -0.053635165,
        -0.027988795,
        0.0028180433,
        -0.038186956,
        0.046643905,
        -0.03817676,
        0.11258302,
        -0.029408582,
        0.007978997,
        0.035129484,
        -0.031633243,
        -0.031122053,
        -0.079740435,
        -0.052736185,
        0.043832034,
        -0.025272286,
        0.055380724,
        -0.0063739363,
        -0.042163342,
        0.031672295,
        -0.019871611,
        0.004979059,
        -0.03099832,
        -0.0399267,
        -0.015902476,
        0.034839842,
        -0.012638347,
        0.012940756,
        -0.61841047,
        0.003959999,
        0.019565048,
        -0.06375712,
        -0.14192313,
        0.024919013,
        0.027794471,
        0.013746951,
        0.06620501,
        0.002184498,
        0.07063507,
        0.010209688,
        -0.01369999,
        0.019723307,
        0.07263453,
        0.08399439,
        0.020995561
      ]
    },
    {
      "id": "worldsfair-speaker-285",
      "conference": "worldsfair",
      "name": "Kate Deyneka",
      "role": "Solo founder",
      "company": "Reelful",
      "twitter": "https://x.com/katedeyneka",
      "talks": [
        {
          "title": "Reelful: AI-generated Reels from photos and clips",
          "description": "AI-powered mobile app that turns photos and short clips into ready-to-post Instagram Reels and TikToks without timeline editing, manual prompting, or voice recording.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.008490469,
        0.034616902,
        0.028796311,
        0.045296647,
        -0.033590384,
        -0.018193705,
        -0.046974152,
        0.025068698,
        0.0129859885,
        -0.22598292,
        -0.002599921,
        -0.004859847,
        -0.040781368,
        -0.030244695,
        -0.06202468,
        0.045600582,
        0.05429537,
        -0.090287186,
        -0.0020577994,
        -0.12913956,
        -0.113611445,
        -0.049630288,
        -0.0052896007,
        0.070944704,
        -0.037926715,
        0.0807872,
        -0.023929927,
        0.01209028,
        -0.11897043,
        0.4620622,
        -0.09875132,
        0.030543927,
        -0.0026853841,
        0.07417321,
        0.048091624,
        0.018068466,
        -0.045578636,
        -0.007029952,
        -0.007878529,
        0.072411925,
        -0.034443203,
        0.074417256,
        -0.042621028,
        0.011755029,
        0.007846321,
        0.024880104,
        -0.032646105,
        -0.014000976,
        0.05250891,
        0.0037225918,
        0.008725803,
        -0.0029212541,
        0.04123185,
        -0.009559937,
        0.08294411,
        0.018321432,
        0.017577969,
        -0.13831611,
        0.036007017,
        -0.044899922,
        -0.09674685,
        0.0030969898,
        0.028722644,
        0.019074615,
        0.022412602,
        -0.1156993,
        0.007979114,
        -0.022271283,
        0.017717382,
        -0.05669688,
        0.0007758629,
        0.02907939,
        -0.044230368,
        0.059971206,
        -0.07795153,
        -0.10859502,
        0.02900259,
        0.02600466,
        0.0098126,
        -0.029696338,
        0.010995849,
        0.040717732,
        -0.083082296,
        -0.030990606,
        -0.004131278,
        -0.030486885,
        0.0024670886,
        0.08169781,
        -0.034864802,
        0.070513085,
        -0.02018549,
        -0.020024208,
        -0.070539676,
        0.02007294,
        -0.0700496,
        0.024183363,
        -0.06140588,
        -0.009388667,
        -0.050322663,
        0.03035611,
        0.014884253,
        -0.02256754,
        -0.021219337,
        -0.05918228,
        -0.09148901,
        -0.03823348,
        0.0042640734,
        0.02008785,
        0.03987786,
        0.02015938,
        -0.027178682,
        -0.6340021,
        -0.049102683,
        0.028875183,
        0.014203236,
        -0.14032057,
        -0.006948024,
        0.017915748,
        0.08336746,
        0.084761806,
        -0.07680784,
        0.065045394,
        -0.032774784,
        0.035030022,
        0.00994059,
        0.02433078,
        0.08161517,
        -0.030428046
      ]
    },
    {
      "id": "worldsfair-speaker-286",
      "conference": "worldsfair",
      "name": "Anil Nadiminti",
      "role": "Senior Solutions Architect",
      "company": "Amazon Web Services",
      "twitter": "https://x.com/super_intel_bot",
      "talks": [
        {
          "title": "When AI Agents Pay and Sellers Monetize: Building x402 Apps for Agentic Commerce on AWS",
          "description": "As Agentic AI moves from chat to execution, autonomous agents need a native way to discover, access, and pay for digital services in real time. This session explores how x402 can turn HTTP into a payment-aware interface for machine-to-machine commerce, unlocking crypto-native patterns like programmable access, pay-per-use APIs, and on-demand monetization for data, tools, and services. We’ll show how to build x402-enabled applications and walk through the architecture, the full agentic payments flow, seller monetization strategies, payment verification, and design tradeoffs involved in making agent-driven transactions secure, scalable, and production-ready. Attendees will leave with practical patterns for building apps where AI agents do not just call APIs — they can discover services, evaluate costs, transact autonomously, and enable new revenue models for sellers.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.009098341,
        0.0827868,
        -0.012150141,
        0.0077777803,
        0.06857203,
        -0.00234534,
        -0.08842384,
        -0.014099997,
        0.04799108,
        -0.20315546,
        -0.024077328,
        -0.0987114,
        0.061411254,
        -0.09406107,
        -0.029327111,
        -0.016441615,
        0.043119688,
        -0.13654022,
        -0.016175887,
        -0.085224286,
        -0.1014084,
        -0.10212566,
        -0.038797375,
        -0.010859829,
        -0.063033074,
        0.10439682,
        0.0066599664,
        0.028737266,
        -0.11704487,
        0.43433574,
        -0.030969,
        0.03239896,
        0.002542265,
        -0.011973105,
        0.063028604,
        -0.073588215,
        -0.049240705,
        -0.05312224,
        0.04465289,
        0.010134766,
        0.0073556746,
        0.044078622,
        -0.07286988,
        0.021248728,
        0.009355132,
        -0.022685677,
        -0.051280443,
        0.012039728,
        -0.0030284103,
        -0.029747354,
        -0.11760203,
        -0.04047449,
        0.09043655,
        0.036343906,
        0.04012346,
        0.039423604,
        0.046409648,
        -0.10612943,
        0.06738117,
        0.027139686,
        -0.09642059,
        -0.0561363,
        -0.059088886,
        0.044836715,
        -0.016591607,
        -0.13821721,
        0.051364392,
        -0.02338371,
        0.011974224,
        -0.05583548,
        0.013471205,
        0.025784435,
        -0.035140704,
        0.071201496,
        -0.026118943,
        -0.124846034,
        -0.07889068,
        0.06995129,
        -0.029787686,
        0.03903846,
        0.009599113,
        0.095103726,
        -0.07657999,
        -0.010422191,
        -0.014408645,
        -0.014816073,
        -0.067555815,
        0.03823781,
        -0.019506834,
        0.043637574,
        0.02079192,
        -0.020442475,
        -0.022594362,
        -0.031012343,
        -0.026487097,
        0.02040495,
        -0.06861244,
        0.036092415,
        0.007269681,
        0.031916764,
        0.016504796,
        -0.04508938,
        0.04729629,
        -0.010847723,
        -0.057189096,
        -0.04514037,
        0.02993418,
        -0.00724246,
        0.08716085,
        -0.01401568,
        0.0133126555,
        -0.6224186,
        0.003309459,
        -0.020031285,
        -0.02206782,
        -0.15609099,
        0.07736459,
        0.012849102,
        0.061546385,
        0.036007456,
        -0.033929083,
        0.06630632,
        -0.03807044,
        0.0053243632,
        0.019105323,
        -0.004451445,
        -2.5890085e-7,
        -0.03146077
      ]
    },
    {
      "id": "worldsfair-speaker-287",
      "conference": "worldsfair",
      "name": "Divakar Kumar",
      "role": "Technical Architect",
      "company": "Flyers Soft Private Limited",
      "twitter": "",
      "talks": [
        {
          "title": "Let's integrate AI Agents in Event-Sourced Systems",
          "description": "Fraud detection has always been a race against time. In traditional event-sourced systems, every transaction, login, or transfer is captured as a sequence of immutable events. These events tell a clear story — but only after the fact. What if events could do more than just record history? What if they could talk back? In this talk, we’ll explore how agentic event-driven systems transform fraud detection. Imagine every PaymentInitiated, LoginAttempt, or DeviceChanged event not just being logged, but immediately consumed by an autonomous Fraud Detection Agent. This agent correlates events across accounts, reasons over historical event streams, and generates new events like SuspiciousActivityFlagged or TransactionHeldForReview. Through a real-world inspired use case in banking and digital payments, we’ll show: - How event sourcing provides the perfect memory layer for fraud detection agents - Patterns for agents to safely inject new domain events without violating invariants - How to avoid runaway feedback loops when multiple agents interact (e.g., fraud + compliance + customer service agents) - Governance, auditing, and explainability challenges when autonomous agents take part in mission-critical workflows By the end of this session, you’ll see how event-driven DDD systems evolve when agents stop being passive consumers and start actively shaping the event stream — turning fraud detection from a reactive process into a proactive, adaptive defense.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.058792323,
        0.032761604,
        -0.0015485054,
        -0.038325436,
        0.05298637,
        -0.045673657,
        -0.077467285,
        0.03457995,
        0.08991347,
        -0.18300617,
        -0.11885182,
        0.008916957,
        -0.03320312,
        -0.054221638,
        -0.04279417,
        -0.051671073,
        0.056319553,
        -0.05992279,
        -0.014057231,
        -0.064974986,
        -0.058225304,
        -0.04406912,
        -0.027759973,
        -0.029842027,
        -0.08344475,
        0.15123144,
        -0.0022707102,
        0.066965945,
        -0.0827037,
        0.44722757,
        -0.01621414,
        0.03369659,
        0.07170684,
        -0.06359113,
        0.005573422,
        -0.019127758,
        -0.029823214,
        -0.023933308,
        0.015346608,
        0.054946825,
        0.001420421,
        0.048660062,
        -0.02819854,
        0.01769074,
        0.016262582,
        -0.023384286,
        -0.0682288,
        0.054302704,
        0.04196441,
        -0.040806297,
        -0.10495996,
        -0.01562811,
        0.003601869,
        -0.064284086,
        0.04272118,
        0.043311495,
        0.097213864,
        -0.12371307,
        0.08001146,
        -0.011427234,
        -0.09147224,
        -0.025651643,
        0.02504185,
        0.03941905,
        0.021753553,
        -0.16861388,
        0.050120976,
        -0.055960685,
        -0.028175091,
        -0.06862344,
        0.0046149385,
        -0.007687332,
        0.0007642771,
        0.07178768,
        -0.052947547,
        -0.1092732,
        -0.05319837,
        0.11286727,
        0.00013958281,
        -0.044956222,
        -0.030059423,
        0.048521943,
        -0.09755251,
        -0.056714717,
        0.027128952,
        -0.015344526,
        -0.05105459,
        0.026236469,
        -0.037055913,
        0.10189072,
        0.06557456,
        -0.058705755,
        -0.057384882,
        0.04309807,
        -0.009419254,
        -0.0022700934,
        -0.05702028,
        0.016673895,
        0.008967891,
        0.056676008,
        0.029211566,
        -0.038096003,
        -0.02296124,
        -0.06443648,
        0.0045315125,
        -0.05210311,
        0.04652434,
        -0.009463147,
        0.083880916,
        -0.017845625,
        0.015193227,
        -0.60570544,
        0.016655594,
        -0.007142926,
        0.014854631,
        -0.14988075,
        0.06159262,
        -0.00065585267,
        0.044266913,
        0.11161959,
        0.012875457,
        0.037215997,
        -0.055389818,
        -0.02158948,
        -0.0019792675,
        0.045249622,
        0.024199499,
        -0.028006377
      ]
    },
    {
      "id": "worldsfair-speaker-288",
      "conference": "worldsfair",
      "name": "Manish Kapur",
      "role": "",
      "company": "Sonar",
      "twitter": "",
      "talks": [
        {
          "title": "Guide, Verify, Solve: The Engineering Discipline Agentic Development Demands",
          "description": "Agentic development is not a productivity story: it's a reliability engineering problem at a scale most teams have never faced. Long-running agent tasks fail at alarming rates, pull requests have grown from 50 lines to 5,000, and cognitive surrender is real—the more capable AI output appears, the less humans interrogate it, right at the moment the stakes are highest. Independent, peer-reviewed research from Carnegie Mellon studying 807 open source projects found that AI agent adoption caused a persistent 30% increase in code analysis warnings and a 41% increase in complexity — with long-term development velocity declining as a result. Agents don't just write code faster, they accumulate debt faster, too. The answer is not to slow agents down, it's to govern and refine the loop they operate inside. Sonar's Agent Centric Development Cycle (AC/DC), defines that loop across three continuous stages: guide agents with project-specific context and constraints before a single line is written; verify rigorously and continuously inside the loop, not downstream in CI; and solve issues automatically before they ever reach a manual review. The deeper insight is that this is not primarily a security story. It's an efficiency story. Codebases riddled with complexity make agents slower, less reliable, and significantly more expensive to run. Every token spent navigating legacy debt is a tax on every future agent run. Well-maintained, low-complexity codebases mean fewer failures, fewer tokens, and faster iteration. The teams that instrument this loop now will do more than ship safely: they'll compound their advantage every time an agent touches their codebase. Verification isn't a cost center. In an agentic world, it's a competitive moat.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.016364131,
        0.023164097,
        0.015158471,
        0.042638145,
        -0.016692312,
        -0.049436335,
        -0.050536137,
        -0.022234788,
        0.04320737,
        -0.22994676,
        -0.056309275,
        -0.013253557,
        0.008816718,
        -0.046987172,
        -0.05484302,
        -0.016615583,
        0.10171115,
        -0.09465041,
        -0.03140592,
        -0.082021005,
        0.043298297,
        -0.060227264,
        0.021365734,
        0.08213735,
        0.02646522,
        0.1590631,
        -0.0037973495,
        0.04178499,
        -0.13800825,
        0.46816325,
        -0.040290724,
        -0.018400187,
        0.08293066,
        -0.065515436,
        0.029076591,
        -0.0023996807,
        0.0046833344,
        -0.0054118894,
        -0.01371079,
        0.032819003,
        -0.08464778,
        0.03302299,
        -0.061177008,
        0.028149657,
        -0.059471466,
        -0.05714063,
        -0.08707721,
        -0.05220032,
        -0.015790878,
        -0.049407482,
        -0.0047804597,
        -0.0053384453,
        0.039544832,
        -0.033038896,
        0.055022098,
        0.016103815,
        -0.031418845,
        -0.0023599996,
        0.060127914,
        -0.073036104,
        -0.08678658,
        0.011745289,
        0.021677319,
        0.023069626,
        -0.023559747,
        -0.08223909,
        0.0805007,
        -0.06638417,
        -0.02108581,
        -0.036526687,
        0.03751078,
        -0.020593034,
        -0.07046603,
        0.03512469,
        -0.076912716,
        -0.11071104,
        -0.02381498,
        0.0738494,
        -0.019161215,
        -0.00009023539,
        -0.020729775,
        0.07147185,
        -0.11913158,
        -0.06576162,
        0.059698034,
        0.0051589385,
        -0.036689576,
        0.037383977,
        -0.029951924,
        0.09266672,
        0.0664456,
        -0.054669492,
        0.03791392,
        0.05446543,
        -0.06754417,
        0.011288997,
        -0.05255952,
        0.02574345,
        -0.029927766,
        0.058083527,
        0.015033537,
        -0.023189785,
        0.07209389,
        0.035651274,
        -0.009079675,
        -0.024301585,
        -0.011823571,
        -0.05471152,
        0.05842773,
        -0.009063568,
        -0.04320311,
        -0.60274184,
        0.026917595,
        -0.043120082,
        -0.06433331,
        -0.14842342,
        0.04332992,
        0.0036988975,
        0.074365735,
        0.08109976,
        0.0108218845,
        0.06672224,
        -0.06520436,
        0.0016519866,
        -0.0023487099,
        0.030081829,
        -0.021273658,
        -0.015409348
      ]
    },
    {
      "id": "worldsfair-speaker-289",
      "conference": "worldsfair",
      "name": "Omri Bruchim",
      "role": "Engineering Group Lead, AI",
      "company": "monday.com",
      "twitter": "https://x.com/omribruchim",
      "talks": [
        {
          "title": "From Systems of Record to Systems of Context",
          "description": "Enterprise AI agents are moving fast, but most of them still hit the same wall in production: they have access to tools, documents, APIs, and databases, but they do not understand the real context of how work gets done. At monday.com, we are building agents that operate across real customer workflows, internal product surfaces, knowledge, permissions, memory, and actions. The hard part is not just calling the right tool or retrieving the right document. The hard part is building a reliable context layer that helps agents understand users, work objects, organizational knowledge, prior decisions, business rules, and the relationships between them. This talk will explore the emerging idea of the context graph: a living, queryable layer that connects entities, history, permissions, decisions, and meaning across an organization. Foundation Capital describes context graphs as the next major enterprise AI opportunity because agents need more than rules. They need decision traces: how rules were applied, where exceptions were made, who approved what, and what precedent actually governs reality. I will share how we think about this opportunity at monday.com, how we are implementing parts of it in practice, and what we have learned from building AI agents inside a real AI work platform. The talk will include concrete examples, including how context is collected, represented, retrieved, governed, and evaluated. The audience will leave with a practical framework for moving beyond one-off RAG pipelines and prompt stuffing toward a reusable context layer that compounds over time, improves agent quality, and becomes a strategic moat for companies building AI-native products.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.027764834,
        0.06669083,
        -0.049950115,
        -0.053499732,
        -0.03890183,
        0.0029374105,
        -0.07090103,
        -0.035181813,
        0.04423351,
        -0.22720742,
        -0.039846946,
        -0.07363708,
        0.052545074,
        -0.012673049,
        -0.05667665,
        -0.0013474524,
        0.07890613,
        -0.0025804583,
        -0.050112758,
        -0.07269259,
        -0.029575493,
        -0.0061059864,
        0.06797792,
        0.045421645,
        -0.04701787,
        0.10417162,
        0.019715808,
        0.08808802,
        -0.13217166,
        0.4581941,
        -0.1323223,
        0.00451579,
        -0.05385674,
        -0.036447696,
        0.09132227,
        -0.073812835,
        -0.10858286,
        -0.04003253,
        0.0068315505,
        0.04200821,
        -0.029762823,
        0.023273014,
        -0.02256866,
        0.04988703,
        -0.029519597,
        -0.08124886,
        -0.01451138,
        0.031216882,
        -0.03904506,
        -0.05398721,
        -0.06165658,
        0.035351284,
        0.06326984,
        0.023030955,
        0.04844592,
        0.0014015011,
        0.08642992,
        -0.050329883,
        0.06057102,
        -0.06640268,
        -0.04209509,
        0.023591746,
        -0.011447454,
        -0.02430249,
        -0.010238569,
        -0.04798283,
        0.061387327,
        -0.088644154,
        -0.022164032,
        -0.066402085,
        0.010487078,
        0.08989363,
        -0.10234691,
        0.05762488,
        -0.044084873,
        -0.09972981,
        -0.043446697,
        0.07616902,
        0.00066961226,
        0.020202044,
        -0.0122368615,
        0.052411143,
        -0.05897513,
        -0.026933106,
        0.019990698,
        -0.036303744,
        -0.11724647,
        0.012404745,
        -0.0037069921,
        0.0790015,
        0.06093275,
        -0.030709539,
        -0.063989826,
        0.009142694,
        -0.051108293,
        -0.0142705785,
        -0.0837984,
        0.050129827,
        -0.0063389614,
        0.050423227,
        0.010768089,
        -0.034974575,
        0.071142025,
        -0.022428665,
        -0.013713606,
        -0.036691193,
        -0.03157569,
        -0.012706777,
        0.11189402,
        0.01859731,
        -0.089205146,
        -0.5901063,
        0.033479534,
        0.0018369415,
        -0.066979714,
        -0.14629017,
        0.066497035,
        -0.016221808,
        0.05062061,
        0.07938971,
        -0.0152804395,
        0.06942556,
        -0.024627239,
        -0.037075073,
        0.0030821804,
        0.03432106,
        0.03957446,
        0.026147397
      ]
    },
    {
      "id": "worldsfair-speaker-290",
      "conference": "worldsfair",
      "name": "Flora Liu",
      "role": "",
      "company": "Notion",
      "twitter": "",
      "talks": [
        {
          "title": "AI in GTM at Notion",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0009717245,
        0.0626504,
        -0.027202683,
        -0.060826283,
        0.011500589,
        -0.02474798,
        -0.033481542,
        0.041354854,
        -0.0008297927,
        -0.20817554,
        -0.016532702,
        -0.0178142,
        0.038253732,
        -0.002945707,
        -0.026716398,
        0.013026395,
        0.053353235,
        -0.062169228,
        -0.049362075,
        -0.09342614,
        -0.057599247,
        -0.093730174,
        0.02427875,
        0.061220802,
        -0.080580845,
        0.074606575,
        0.07280004,
        -0.0127065275,
        -0.14830963,
        0.4588607,
        -0.039647542,
        0.0082228305,
        -0.005817248,
        0.004563554,
        0.015846435,
        -0.051374584,
        -0.042450793,
        -0.0034019882,
        0.024263414,
        0.04880041,
        0.02492839,
        0.038699947,
        -0.0127068665,
        0.04612995,
        -0.07689987,
        -0.0063308924,
        -0.10630228,
        -0.009724602,
        -0.0088276,
        0.007394874,
        -0.04914644,
        0.0022707055,
        0.01097247,
        -0.05847695,
        0.0014403266,
        0.06700951,
        -0.04115995,
        -0.04413947,
        0.16335487,
        -0.0040914537,
        -0.19255245,
        -0.021171203,
        -0.03691936,
        0.015530431,
        0.022051018,
        -0.011452898,
        0.031828947,
        -0.07170856,
        0.002904593,
        -0.04206469,
        -0.0013113315,
        0.06486937,
        -0.008765368,
        0.028497277,
        -0.062843286,
        -0.07544585,
        0.034936216,
        0.074011475,
        -0.029741898,
        -0.05427335,
        0.044528306,
        0.08672689,
        -0.16196565,
        -0.06520969,
        -0.007697279,
        -0.0014191172,
        -0.045503106,
        0.042230766,
        -0.035847705,
        0.0849993,
        0.05410161,
        -0.0148958415,
        0.022872133,
        0.0118586365,
        -0.10736564,
        0.058761448,
        -0.024249025,
        0.057863366,
        -0.044361375,
        0.022226747,
        -0.033499885,
        -0.050024323,
        0.028426811,
        0.010985472,
        -0.094600186,
        -0.030878868,
        0.013928464,
        -0.0051023536,
        0.013233664,
        0.028784975,
        -0.016344812,
        -0.6079243,
        0.009302549,
        -0.041770358,
        -0.0047714454,
        -0.1505582,
        -0.012707745,
        -0.031266585,
        0.055564404,
        0.0927292,
        -0.033232335,
        0.017125187,
        0.03852931,
        -0.039836638,
        0.034308814,
        0.058849607,
        0.052643925,
        0.0056838635
      ]
    },
    {
      "id": "worldsfair-speaker-291",
      "conference": "worldsfair",
      "name": "Jared Joselowitz",
      "role": "Research engineer",
      "company": "Ufonia",
      "twitter": "",
      "talks": [
        {
          "title": "Shipping AI to a Million Patients Without an A/B Test",
          "description": "You can't A/B test on patients. You can't unsend a phone call. The model card won't save you at the post-incident review. Most AI eng playbooks assume the opposite. Ship to 5%, watch the dashboard, roll back if it goes wrong. None of it survives regulated deployment, which is now coming for fintech, legal, and government too. So the engineering has to move: into hazard analysis, simulated populations, asymmetric evaluation, and audit trails treated as the deliverable. The trail is the product. I'll show you what changes when rollback isn't an option. How Ufonia ships Dora, an AI voice agent now making clinical follow-up calls on the NHS and across US health systems, using a hazard-driven simulation rig (MATRIX) and a prompt-optimisation flywheel that surface failures and conform the same base system to each clinical niche, all of it pinned to an audit trail. And the cheap version of all this, for any team whose users can't be the test population.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0045243218,
        0.047918394,
        0.061215416,
        0.0025931885,
        0.012362622,
        -0.0029580656,
        -0.067665696,
        -0.026538027,
        0.030072574,
        -0.20958513,
        -0.0447904,
        -0.021220755,
        0.051346414,
        -0.0640005,
        -0.04583114,
        0.07925764,
        0.121108584,
        -0.042588905,
        -0.02008145,
        -0.079432145,
        -0.013423472,
        -0.04968866,
        -0.018170381,
        0.08802336,
        -0.052430708,
        0.07808565,
        0.031019606,
        0.024871286,
        -0.095026955,
        0.4581487,
        -0.11295429,
        -0.017812887,
        0.025049113,
        -0.0011733939,
        0.056697857,
        -0.03110783,
        0.0006217973,
        0.011196309,
        -0.030588843,
        0.04008761,
        0.041024264,
        0.005509158,
        -0.028874172,
        0.054961763,
        -0.0067875055,
        -0.062850356,
        -0.054451086,
        0.03185425,
        0.0060418677,
        -0.005886766,
        -0.08920129,
        -0.0051368997,
        0.045496643,
        -0.014789048,
        0.020895608,
        -0.016185716,
        0.036547773,
        -0.10849126,
        0.08767389,
        -0.015103871,
        -0.07266642,
        0.04047764,
        -0.016423993,
        -0.03110757,
        0.026550092,
        -0.095617324,
        0.085256904,
        -0.037438806,
        -0.022501541,
        0.040535163,
        -0.01884103,
        0.036898904,
        -0.09999493,
        -0.013436892,
        -0.09516542,
        -0.09551169,
        -0.042193025,
        0.07256279,
        0.029332656,
        0.024477895,
        0.026936274,
        0.09376351,
        -0.057626553,
        0.047321726,
        0.023396133,
        0.0007990702,
        -0.06941989,
        0.08390249,
        0.005526939,
        0.05742621,
        0.0045596384,
        -0.042139746,
        -0.06078318,
        0.040540848,
        -0.024587465,
        -0.034902822,
        -0.019211652,
        0.02148901,
        0.03210675,
        -0.01878472,
        0.00614913,
        -0.046354406,
        0.051694788,
        0.035089184,
        -0.04047246,
        -0.11268622,
        -0.059396666,
        0.033470202,
        0.042477254,
        0.031553302,
        -0.0018543439,
        -0.6262824,
        0.00064380973,
        0.0039419406,
        -0.044761445,
        -0.16835105,
        0.07114943,
        0.018379955,
        -0.0099911,
        0.119476415,
        0.02285648,
        -0.015020152,
        -0.014970479,
        0.0404853,
        -0.024898294,
        0.077832885,
        0.08616108,
        0.017254343
      ]
    },
    {
      "id": "worldsfair-speaker-292",
      "conference": "worldsfair",
      "name": "Ashok Chandrasekar",
      "role": "",
      "company": "Google",
      "twitter": "",
      "talks": [
        {
          "title": "Are LLM Performance Benchmarks Reliable?",
          "description": "Standardizing performance benchmarks for production-grade Large Language Models is currently a significant challenge across the industry. Conflicting data is prevalent, whether originating from server developers like vLLM and SGLang or from various analysts and competitive benchmarks, and these results often fail to hold up under real-world conditions. Our research into these inconsistencies identified several critical factors, including the constraints of single-process tools, specifically the Python Global Interpreter Lock (GIL) and the nuances of model-level settings like temperature. Furthermore, a lack of transparency regarding load generation parameters such as QPS and concurrency, paired with insufficient observability into the benchmarking clients themselves, contributes to these disparate outcomes. In this talk, we share key lessons learned from our benchmarking efforts, examining the primary pitfalls that distort performance data and offering strategies for mitigation. Additionally, we will introduce Inference Perf, an open-source, multi-process utility we developed to provide reliable stress-testing for production stacks. Our goal is to promote standardized, real-world benchmarking practices that allow the community to move beyond unreliable data. Join us to discover how to accurately measure, optimize, and report LLM performance with certainty.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.1644743,
        0.104011744,
        0.013773105,
        0.045106184,
        0.056512658,
        -0.047724783,
        -0.07505211,
        -0.002502063,
        0.012509047,
        -0.23835813,
        0.006663897,
        -0.0058102016,
        0.029407233,
        -0.030866979,
        -0.020172445,
        0.03434638,
        0.06430401,
        -0.06615194,
        -0.10224906,
        -0.0702289,
        -0.020453855,
        -0.08552977,
        0.007682092,
        0.017345594,
        -0.082200624,
        0.067692,
        0.012000826,
        0.019770054,
        -0.1211251,
        0.43509647,
        -0.12675384,
        -0.022541272,
        0.0022768478,
        -0.1401959,
        0.030372389,
        -0.05803067,
        0.028935425,
        0.07891118,
        0.033702422,
        0.021598024,
        0.032775845,
        0.016953418,
        -0.06852823,
        0.08666379,
        -0.02402753,
        -0.03616212,
        -0.04064989,
        0.04673581,
        0.003624446,
        -0.0075952606,
        -0.04054907,
        -0.057438053,
        0.04238395,
        0.005052943,
        -0.009169123,
        0.022830857,
        -0.040143147,
        -0.01228931,
        0.16165157,
        -0.043873467,
        -0.11382926,
        0.0014788944,
        0.02432121,
        -0.011721106,
        -0.001710671,
        -0.025897726,
        0.057573684,
        -0.04132248,
        -0.014430553,
        0.02100416,
        0.02836556,
        -0.028559413,
        -0.10915581,
        -0.0037228558,
        -0.025478153,
        -0.10835027,
        -0.08480912,
        0.06956196,
        -0.04326583,
        0.03413157,
        0.03391911,
        0.08839691,
        -0.03500418,
        -0.04524057,
        0.06526343,
        0.039624184,
        -0.026679361,
        0.06058049,
        -0.052611824,
        0.13494278,
        0.01601446,
        -0.021474682,
        -0.11284799,
        -0.055539712,
        -0.019109517,
        0.06092549,
        -0.08618754,
        0.028393792,
        -0.015477852,
        0.050929833,
        0.031642683,
        -0.031103346,
        0.03209456,
        0.001988216,
        0.04532331,
        -0.05915173,
        0.021968108,
        0.00016243458,
        -0.00975197,
        -0.082554966,
        -0.06203277,
        -0.566814,
        0.049065832,
        0.03534825,
        -0.049666937,
        -0.14575106,
        0.05892663,
        -0.009344363,
        -0.030009577,
        0.11370637,
        -0.02232273,
        0.09114723,
        -0.02632272,
        0.037644863,
        -0.012329061,
        0.021680871,
        0.032229844,
        -0.014094316
      ]
    },
    {
      "id": "worldsfair-speaker-293",
      "conference": "worldsfair",
      "name": "Jason Kramberger",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Are LLM Performance Benchmarks Reliable?",
          "description": "Standardizing performance benchmarks for production-grade Large Language Models is currently a significant challenge across the industry. Conflicting data is prevalent, whether originating from server developers like vLLM and SGLang or from various analysts and competitive benchmarks, and these results often fail to hold up under real-world conditions. Our research into these inconsistencies identified several critical factors, including the constraints of single-process tools, specifically the Python Global Interpreter Lock (GIL) and the nuances of model-level settings like temperature. Furthermore, a lack of transparency regarding load generation parameters such as QPS and concurrency, paired with insufficient observability into the benchmarking clients themselves, contributes to these disparate outcomes. In this talk, we share key lessons learned from our benchmarking efforts, examining the primary pitfalls that distort performance data and offering strategies for mitigation. Additionally, we will introduce Inference Perf, an open-source, multi-process utility we developed to provide reliable stress-testing for production stacks. Our goal is to promote standardized, real-world benchmarking practices that allow the community to move beyond unreliable data. Join us to discover how to accurately measure, optimize, and report LLM performance with certainty.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.10394653,
        0.057151206,
        0.04803469,
        0.037802875,
        0.04343118,
        -0.0624642,
        -0.04074719,
        0.0035283116,
        0.03134224,
        -0.18816812,
        -0.04806938,
        0.003863597,
        0.007967551,
        0.0050651864,
        -0.07228918,
        0.053409483,
        0.115677,
        -0.051001344,
        -0.107707255,
        -0.08560509,
        -0.04876043,
        -0.07149924,
        0.08777084,
        0.050247673,
        -0.052930783,
        0.12749575,
        0.020346165,
        0.020099945,
        -0.13278654,
        0.47008365,
        -0.089757696,
        0.0056843567,
        -0.029876301,
        -0.09521385,
        0.03230051,
        -0.06831066,
        0.02593402,
        0.044428468,
        0.046458833,
        0.03951993,
        0.05401401,
        0.04315401,
        -0.06359631,
        0.079513974,
        -0.02922842,
        0.010199428,
        -0.03424974,
        0.04652976,
        0.040906932,
        -0.03446041,
        -0.051154286,
        -0.02568362,
        0.053123858,
        -0.001453961,
        -0.014744037,
        0.02487897,
        0.034797642,
        0.0081488425,
        0.16184027,
        -0.048228912,
        -0.06773384,
        -0.039214224,
        -0.021108452,
        -0.0022876211,
        0.037308846,
        -0.036678337,
        0.05554151,
        -0.06435061,
        -0.014200343,
        0.012296972,
        0.042406775,
        0.023920517,
        -0.077602856,
        0.016870188,
        -0.08139692,
        -0.0857227,
        -0.06524801,
        0.07952333,
        -0.029925141,
        0.0066139014,
        -0.012076598,
        0.09563797,
        0.01760106,
        -0.037185635,
        0.06498704,
        0.08345193,
        -0.08215874,
        0.062220912,
        -0.01929968,
        0.12495173,
        0.079769775,
        -0.016688606,
        -0.048143916,
        -0.026912099,
        -0.0093184365,
        0.034532066,
        -0.069381125,
        -0.013695039,
        -0.00048778232,
        0.054954104,
        -0.0049672886,
        0.0047447947,
        0.0057914136,
        -0.036654953,
        0.02433611,
        -0.023644645,
        -0.010208954,
        -0.0046131765,
        -0.025796048,
        0.00871606,
        -0.03134524,
        -0.60370815,
        0.02694031,
        -0.019890126,
        -0.047908764,
        -0.13981108,
        0.040591013,
        0.056311797,
        -0.02660676,
        0.036275685,
        0.026518237,
        0.05878693,
        0.0037549655,
        -0.0051728664,
        -0.033728704,
        0.0664501,
        0.009897397,
        0.01412487
      ]
    },
    {
      "id": "worldsfair-speaker-294",
      "conference": "worldsfair",
      "name": "Daniel Kim",
      "role": "Head of Growth",
      "company": "Cerebras Systems",
      "twitter": "https://x.com/learnwdaniel",
      "talks": [
        {
          "title": "All the Things We Have to Do to Satisfy Your Insatiable Need for Tokens",
          "description": "Every time the industry figures out how to serve tokens faster and cheaper, the appetite grows to match. Models get bigger, contexts get longer, agents start chaining thousands of calls together. The finish line keeps moving. This talk is a technical tour through everything the industry has done to keep up, led by two experts in high-performance inference. We'll start with the optimizations that made hardware work harder without changing the underlying architecture. Then we'll go up a level with techniques that work smarter across requests and across the model itself. And finally, a peek into the future with heterogeneous disaggregated inference, the architectural shift that splits prefill and decode across specialized hardware, and even more advanced forms of hardware specialization coming your way soon. Token demand is about to get a lot more insatiable. Let's see what the future has in store for us!",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.07439365,
        0.089331836,
        -0.026860649,
        0.0067715016,
        0.017092265,
        -0.06631152,
        -0.038704492,
        -0.026170898,
        0.053106364,
        -0.2017464,
        -0.07808404,
        -0.07828657,
        0.0107559245,
        -0.037511952,
        -0.09752369,
        -0.03134895,
        0.06627557,
        -0.10560794,
        -0.0730269,
        -0.021239853,
        -0.036332622,
        -0.053386945,
        0.025618883,
        0.08370515,
        -0.0446689,
        0.07860747,
        0.09454693,
        0.038536407,
        -0.09084138,
        0.43099877,
        -0.12703326,
        0.02344668,
        -0.02186877,
        0.008366641,
        0.049683277,
        -0.0497125,
        -0.05971619,
        -0.030602258,
        -0.013039706,
        0.056012664,
        0.0059424164,
        0.031708572,
        -0.022508997,
        0.075235255,
        -0.05757813,
        -0.039702155,
        -0.088748544,
        0.03783894,
        0.007849652,
        -0.011694394,
        -0.049348164,
        0.03794423,
        0.022164946,
        0.0412959,
        0.06300281,
        0.036354292,
        0.052551407,
        -0.05769197,
        0.13640921,
        -0.058164414,
        -0.13593394,
        0.09715789,
        -0.032857843,
        -0.0026653495,
        0.010836531,
        -0.040785298,
        0.011889027,
        -0.008172098,
        -0.011490069,
        -0.016706688,
        0.027345877,
        0.005504658,
        -0.07085083,
        0.010967752,
        -0.07497701,
        -0.112860024,
        -0.01808998,
        0.07421815,
        -0.053278953,
        -0.011959157,
        0.038096484,
        0.057313085,
        -0.007964467,
        -0.074181125,
        0.1066075,
        -0.05866684,
        -0.015234351,
        -0.0069610854,
        -0.0046503567,
        0.09922817,
        0.06663288,
        0.033353854,
        -0.107862145,
        -0.0047922283,
        -0.100886606,
        0.00027461097,
        -0.07257758,
        0.0090802945,
        -0.0011508301,
        0.045984562,
        0.030575244,
        -0.006241656,
        0.011899856,
        -0.003069561,
        -0.025213992,
        -0.029498592,
        -0.026755994,
        -0.026383027,
        0.080986366,
        0.004878752,
        -0.0522492,
        -0.6233718,
        0.04097741,
        -0.08072036,
        -0.008384163,
        -0.119969495,
        0.026848579,
        -0.05047895,
        0.0660201,
        0.023618259,
        -0.06659168,
        0.008520317,
        0.00041907435,
        -0.0033637574,
        -0.022904549,
        0.025768107,
        0.025034096,
        0.023211034
      ]
    },
    {
      "id": "worldsfair-speaker-295",
      "conference": "worldsfair",
      "name": "Natalie Serrino",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "All the Things We Have to Do to Satisfy Your Insatiable Need for Tokens",
          "description": "Every time the industry figures out how to serve tokens faster and cheaper, the appetite grows to match. Models get bigger, contexts get longer, agents start chaining thousands of calls together. The finish line keeps moving. This talk is a technical tour through everything the industry has done to keep up, led by two experts in high-performance inference. We'll start with the optimizations that made hardware work harder without changing the underlying architecture. Then we'll go up a level with techniques that work smarter across requests and across the model itself. And finally, a peek into the future with heterogeneous disaggregated inference, the architectural shift that splits prefill and decode across specialized hardware, and even more advanced forms of hardware specialization coming your way soon. Token demand is about to get a lot more insatiable. Let's see what the future has in store for us!",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.1016203,
        0.08290218,
        0.031111572,
        0.045724235,
        0.09879574,
        -0.073157154,
        -0.082807116,
        0.022519464,
        0.04200089,
        -0.15968552,
        0.032244436,
        0.014779331,
        -0.02114971,
        -0.031087192,
        -0.0770544,
        -0.003744748,
        0.12978661,
        -0.101083405,
        -0.052264974,
        -0.03083224,
        0.024959225,
        -0.11410362,
        0.039627787,
        0.022559097,
        -0.05717048,
        0.10348134,
        0.019781053,
        -0.0014998917,
        -0.103617825,
        0.4433701,
        -0.07508499,
        0.016186697,
        0.00878587,
        -0.04510901,
        0.037917998,
        -0.08201907,
        -0.009270285,
        -0.037867755,
        -0.030955119,
        0.07662493,
        0.09012873,
        0.05883227,
        -0.071209945,
        0.04871003,
        -0.03177253,
        -0.017750517,
        -0.090191275,
        0.0145770805,
        -0.0024599412,
        -0.028284619,
        -0.033508856,
        0.008874226,
        -0.0010168866,
        0.023024032,
        0.028431198,
        -0.010353572,
        -0.013177462,
        -0.080870524,
        0.14120999,
        -0.075405635,
        -0.09195421,
        0.090552166,
        -0.038655248,
        -0.024996713,
        0.035812702,
        -0.046129595,
        0.03827259,
        -0.019927543,
        -0.01639553,
        -0.03374356,
        0.016981613,
        0.010729909,
        -0.063011244,
        0.013502561,
        0.004867598,
        -0.18686302,
        -0.059298225,
        0.031104201,
        0.016908186,
        0.023800805,
        -0.005218616,
        0.040630117,
        -0.045222994,
        -0.029343905,
        0.044954017,
        -0.037226778,
        -0.023634493,
        0.064059936,
        -0.08293549,
        0.101276726,
        0.047325004,
        0.024125198,
        -0.058213722,
        0.00948557,
        -0.034449004,
        -0.03678351,
        -0.06182595,
        0.05030494,
        -0.0028994225,
        0.0201709,
        0.021501558,
        -0.04251893,
        0.013472667,
        0.017567797,
        -0.036554273,
        -0.048922382,
        -0.08883505,
        -0.034385517,
        0.06738716,
        0.06777338,
        -0.04116194,
        -0.60595584,
        0.07314921,
        -0.06140647,
        -0.05524637,
        -0.14903854,
        0.02456172,
        0.0210787,
        0.0059780558,
        0.07717212,
        -0.025593653,
        0.037633106,
        -0.06059141,
        0.000825481,
        0.037387524,
        -0.008926482,
        0.07829513,
        -0.013096179
      ]
    },
    {
      "id": "worldsfair-speaker-296",
      "conference": "worldsfair",
      "name": "Khaled Alashmouny",
      "role": "",
      "company": "AIDAChip",
      "twitter": "",
      "talks": [
        {
          "title": "What If Your Chip Design Team Moved Like a Single Body?",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0055468096,
        0.040117625,
        -0.083111994,
        0.0051020123,
        -0.04566923,
        0.03622443,
        -0.026867047,
        -0.013849947,
        0.07053493,
        -0.22040136,
        0.0033348904,
        -0.057137925,
        -0.025540581,
        -0.051205795,
        -0.03145623,
        -0.004941863,
        0.112498805,
        -0.09676143,
        -0.055648334,
        -0.05274206,
        -0.025837949,
        -0.0394971,
        -0.026900392,
        0.027120281,
        -0.10131401,
        0.101010375,
        0.013880658,
        0.037088472,
        -0.10091769,
        0.49291262,
        -0.1171669,
        -0.046697278,
        -0.017288469,
        -0.08933839,
        0.030690439,
        0.013594086,
        -0.010606422,
        0.01461257,
        0.06323399,
        0.11754901,
        -0.0067364564,
        0.05032131,
        -0.07756709,
        0.035694532,
        0.04105596,
        -0.068496026,
        -0.064950615,
        -0.045371357,
        -0.015795665,
        -0.016417144,
        0.0004550092,
        0.0068997494,
        0.05085625,
        -0.035120565,
        0.040694606,
        0.061044782,
        0.035746247,
        0.040140644,
        0.028240094,
        -0.06387476,
        -0.05960797,
        -0.03750053,
        0.020547362,
        0.039907265,
        -0.012205514,
        0.002477741,
        0.0588058,
        -0.059637353,
        -0.007154928,
        0.00083618145,
        0.03342333,
        0.025483916,
        -0.029741446,
        -0.008479492,
        0.0007215077,
        -0.011023599,
        0.054093167,
        0.04924176,
        -0.006174504,
        -0.008665489,
        0.10538291,
        0.025877643,
        -0.04533076,
        -0.030277943,
        -0.018615585,
        -0.011542481,
        -0.10677632,
        0.09740534,
        -0.0052400576,
        0.023586892,
        0.018118108,
        -0.00196214,
        0.017931875,
        0.028416602,
        -0.042571608,
        0.02229766,
        -0.006940651,
        0.038726952,
        0.037618138,
        -0.074406326,
        -0.013451808,
        -0.00086589664,
        0.0737571,
        0.0041925637,
        0.055720896,
        -0.114835076,
        -0.022670873,
        -0.026861893,
        0.13685824,
        -0.038739413,
        -0.028655902,
        -0.60568434,
        -0.0021442068,
        -0.016472127,
        0.00953728,
        -0.15954123,
        -0.05857959,
        -0.03509205,
        0.0008757299,
        0.10583831,
        -0.015058474,
        0.049957756,
        -0.06073902,
        -0.036790237,
        0.019777486,
        0.039065085,
        -0.0122040175,
        -0.040630102
      ]
    },
    {
      "id": "worldsfair-speaker-297",
      "conference": "worldsfair",
      "name": "Miguel González Fernández",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Art of Building Verifiers for Computer Use Agents",
          "description": "Every team building browser agents has the same problem: you can't trust your own evals. Browser tasks are too open-ended for deterministic checks, so teams use LLM verifiers as judges, and the judges are wrong constantly. WebVoyager misses 45% of failures. WebJudge misses 22%. Used as RL reward, you're not training a better agent, you're training a more confident liar. This talk walks through the Universal Verifier, open-sourced with Microsoft Research: false positive rate near zero, Cohen's kappa matching human-human agreement. Four design principles, one open benchmark, and an honest account of where auto-research worked and where it plateaued.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.07784842,
        0.00791819,
        0.018359557,
        0.03936095,
        0.005398089,
        -0.048524126,
        -0.07433047,
        -0.0004815191,
        -0.017492127,
        -0.26316664,
        -0.025159484,
        0.0051330603,
        0.005282051,
        -0.027928663,
        -0.05984253,
        -0.0032538003,
        0.0990622,
        -0.061187837,
        -0.051309135,
        -0.052985433,
        -0.08914237,
        -0.05558241,
        -0.0035002474,
        0.06889788,
        0.01634468,
        0.070985444,
        0.0378694,
        0.05881691,
        -0.07394348,
        0.4613549,
        -0.05361651,
        0.017569555,
        0.02684197,
        -0.00042035052,
        -0.011749374,
        -0.07978215,
        0.023151923,
        0.024496976,
        0.069115415,
        0.038640145,
        0.016449597,
        0.09999048,
        -0.028731091,
        0.016651245,
        0.01356474,
        -0.0659238,
        -0.026332214,
        0.024236605,
        -0.038889766,
        -0.04864478,
        -0.055917624,
        0.014591761,
        0.019459408,
        -0.053672496,
        -0.009369176,
        0.01179776,
        0.073874496,
        -0.033934552,
        0.07629948,
        -0.010622034,
        -0.017199354,
        -0.031550616,
        0.032880753,
        0.030605966,
        0.022994852,
        -0.04615323,
        0.0068715387,
        -0.0995919,
        0.006433868,
        -0.051875502,
        0.010667002,
        -0.037437074,
        0.021000562,
        0.019781733,
        -0.0134303905,
        -0.13301061,
        -0.104301445,
        0.009259088,
        -0.051926132,
        0.021755086,
        -0.04164151,
        0.03730995,
        -0.01916078,
        -0.020575188,
        0.06682679,
        -0.014462128,
        -0.062472656,
        0.07157233,
        -0.08440834,
        0.06598116,
        0.016781868,
        -0.000030136884,
        -0.089988746,
        0.003322366,
        -0.02840505,
        -0.019331118,
        -0.030904714,
        0.061911717,
        -0.027675701,
        0.017073488,
        -0.055642486,
        0.026387384,
        0.04224256,
        0.031741243,
        0.010255408,
        -0.12760909,
        0.01088964,
        0.030041011,
        0.06980645,
        -0.04680013,
        -0.022737678,
        -0.63111544,
        -0.03315576,
        -0.021690536,
        -0.028932191,
        -0.18380086,
        0.06542227,
        0.0288949,
        0.023531381,
        0.0861997,
        0.011962272,
        0.071577705,
        -0.0072675957,
        0.02697112,
        -0.04739798,
        0.009352565,
        0.05202143,
        0.010901619
      ]
    },
    {
      "id": "worldsfair-speaker-298",
      "conference": "worldsfair",
      "name": "Corby Rosset",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Art of Building Verifiers for Computer Use Agents",
          "description": "Every team building browser agents has the same problem: you can't trust your own evals. Browser tasks are too open-ended for deterministic checks, so teams use LLM verifiers as judges, and the judges are wrong constantly. WebVoyager misses 45% of failures. WebJudge misses 22%. Used as RL reward, you're not training a better agent, you're training a more confident liar. This talk walks through the Universal Verifier, open-sourced with Microsoft Research: false positive rate near zero, Cohen's kappa matching human-human agreement. Four design principles, one open benchmark, and an honest account of where auto-research worked and where it plateaued.",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05983356,
        0.06412583,
        0.059601128,
        0.03375393,
        -0.0013972562,
        -0.042751342,
        -0.09681794,
        -0.0039237854,
        -0.022495234,
        -0.22491872,
        -0.04552092,
        -0.016316613,
        0.046324532,
        -0.022650048,
        -0.060311366,
        0.0128051415,
        0.09774298,
        -0.06965252,
        -0.07486135,
        -0.08816747,
        -0.026626289,
        -0.06697006,
        -0.018344598,
        0.106676705,
        0.013164911,
        0.083061956,
        0.027851906,
        0.06879236,
        -0.10125578,
        0.4692562,
        -0.04654424,
        0.017488055,
        -0.0019321867,
        0.00826734,
        0.041582193,
        -0.06664164,
        0.011855977,
        -0.016318306,
        0.038836457,
        0.015452247,
        0.049714595,
        0.034666117,
        -0.019120583,
        0.025200525,
        0.006937321,
        -0.062364146,
        -0.05555254,
        -0.0071148043,
        -0.018825978,
        -0.048885353,
        -0.033700008,
        -0.022661475,
        -0.012471468,
        -0.075816594,
        0.022502543,
        0.029443584,
        0.053017836,
        -0.055697717,
        0.08879566,
        -0.008920375,
        -0.0894233,
        -0.013401524,
        -0.03491194,
        -0.015571312,
        0.0029696638,
        -0.08043383,
        0.004673839,
        -0.04815123,
        0.005254327,
        -0.012697674,
        0.009053695,
        0.026627576,
        -0.018884886,
        0.03389986,
        -0.08716888,
        -0.09657207,
        -0.10578552,
        0.0361948,
        -0.041938197,
        0.028730448,
        -0.00526608,
        0.036280584,
        -0.034417655,
        -0.040612496,
        0.046386935,
        -0.003800532,
        -0.051460475,
        0.041600395,
        -0.08594474,
        0.059053753,
        0.039259985,
        -0.029819762,
        -0.072701894,
        0.013319118,
        -0.039460577,
        -0.028496688,
        -0.06579642,
        0.022931632,
        -0.024052693,
        -0.0007219843,
        -0.03502239,
        0.020791259,
        0.070645265,
        0.01344685,
        0.027017578,
        -0.11845851,
        -0.0150024,
        0.03617216,
        0.05686779,
        -0.09140975,
        -0.0580013,
        -0.6244574,
        -0.058357637,
        0.0055777547,
        -0.06467894,
        -0.16560255,
        0.06043497,
        0.0031628928,
        0.090460226,
        0.08195522,
        -0.007782624,
        0.04670089,
        -0.001989868,
        -0.00032119654,
        -0.01613873,
        0.043450665,
        0.065146685,
        0.009764982
      ]
    },
    {
      "id": "worldsfair-speaker-299",
      "conference": "worldsfair",
      "name": "Sheilah Kirui",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Seeing the Plumbing: Profiling vLLM Speculative Decoding on NVIDIA Blackwell",
          "description": "Speculative decoding promises dramatic LLM speedups by using a tiny draft model to guess tokens ahead of a large target model. However, dual-model serving fundamentally rewrites your memory dynamics and introduces a rigid engineering trade-off: guess right, and you bypass the memory-bandwidth bottleneck; guess wrong, and you waste compute. This session is a live-demo routing identical workloads through baseline and speculative configurations in vLLM on a single NVIDIA RTX 6000 Blackwell GPU. Splitting the screen between a Streamlit app and a live Grafana dashboard, we will profile the inference engine across three vectors: Time per Output Token (TPOT): The real-time, user-facing latency delta. KV Cache & Memory Footprint: The exact VRAM tax of tracking parallel token states within a 96GB budget. Draft Acceptance Rate: Visualizing the tipping point where dropping acceptance rates cause speculative decoding to fall below baseline efficiency.Supporting MaterialsProject Repository: https://github.com/akamai-developers/speculative-decoding-example-vllm-blackwell# (Work In Progress / Active Development)",
          "day": "Day 4 — Session Day 3",
          "time": "11:40am-12:00pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.088188544,
        0.07028875,
        0.076007605,
        0.057878915,
        0.021064505,
        -0.021042943,
        -0.049740925,
        -0.027535835,
        0.066334225,
        -0.23169881,
        -0.028702904,
        -0.031158928,
        -0.06439575,
        -0.016490074,
        -0.0337343,
        0.037866823,
        0.074821636,
        -0.05815076,
        -0.03690651,
        -0.034854453,
        -0.044718143,
        -0.09576559,
        0.027474262,
        0.050774634,
        -0.10634414,
        0.034197982,
        -0.016393274,
        0.043849163,
        -0.10988773,
        0.46098456,
        -0.09277579,
        -0.03979851,
        -0.010233931,
        -0.04584077,
        0.028168324,
        -0.07764059,
        0.040688332,
        0.058753904,
        0.047212247,
        0.07095453,
        0.039983667,
        -0.037688214,
        -0.063190915,
        0.06448508,
        -0.050211992,
        -0.049483173,
        -0.050502762,
        0.025746463,
        0.002840785,
        -0.00045574384,
        -0.024407115,
        -0.04775319,
        0.05474536,
        0.02193964,
        0.026711738,
        0.080523625,
        -0.034535907,
        -0.09177855,
        0.12106504,
        -0.026463712,
        -0.12617333,
        -0.008506999,
        0.031722967,
        0.057959672,
        0.01635033,
        -0.0797489,
        0.04686662,
        -0.06594859,
        -0.0073986067,
        -0.0077475626,
        -0.015506647,
        -0.044367872,
        -0.048538614,
        0.049607877,
        -0.0028534192,
        -0.15146269,
        -0.035416484,
        0.031680193,
        -0.01154222,
        -0.019210188,
        -0.01681608,
        0.09641648,
        -0.049111307,
        0.006356581,
        -0.030868357,
        0.009763582,
        -0.08658633,
        0.041230228,
        -0.024169443,
        0.047463007,
        0.031683184,
        -0.02377903,
        -0.05874968,
        0.0058446354,
        -0.008948919,
        0.032146506,
        -0.013443144,
        0.08748822,
        -0.010888161,
        0.038750052,
        0.030098317,
        -0.0027396297,
        0.035135828,
        -0.023484893,
        0.019775094,
        -0.05301385,
        -0.023720633,
        -0.007905161,
        0.08197225,
        0.001872876,
        -0.01973888,
        -0.62128377,
        -0.027423998,
        -0.0021260856,
        -0.07592605,
        -0.14665638,
        0.033083186,
        -0.0032700512,
        0.044522278,
        0.07928377,
        0.017201561,
        0.05872629,
        0.006208596,
        -0.05548012,
        -0.009810461,
        0.01912226,
        0.021746619,
        0.020676974
      ]
    },
    {
      "id": "worldsfair-speaker-300",
      "conference": "worldsfair",
      "name": "Mike Chambers",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Harness Engineering: Building the Production Cage for Powerful Domain Agents",
          "description": "Every agent is a while loop. The model takes strings in and produces strings out. We've all written it, debugged it, shipped it. And yet every team building agents is still re-inventing the same session management, truncation logic, tool wiring, and memory plumbing from scratch. The hard part is the harness: session isolation, context management, memory persistence, sandboxed execution, observability. The machinery that makes a model dependable in production. Most of the failures we see in deployed agents (context rot, premature completion, tool bloat) trace back to harness problems, not model problems. This talk covers what a harness actually does, why \"harness engineering\" suddenly showed up in engineering posts from everyone, and what changes when you stop building harnesses by hand. In live demos, we'll build the same agent three ways: hand-rolled Python, framework-generated, and fully managed through a single API call. Each level shifts the failure modes from infrastructure plumbing to engineering judgment, where the real questions are what context to preserve, when to verify, and how to keep an agent from finishing half the job and calling it done. The harness handles the machinery. You still have to engineer the behavior.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.050166775,
        0.07451597,
        0.080260724,
        0.042433567,
        -0.00037297167,
        -0.029182209,
        -0.048839033,
        -0.016737644,
        0.02839671,
        -0.22725354,
        -0.09231391,
        -0.021907201,
        0.010803151,
        -0.03291601,
        -0.038830794,
        0.009003411,
        0.12176186,
        -0.0768329,
        -0.06216642,
        -0.10782585,
        -0.07476352,
        -0.09539097,
        -0.005395488,
        0.013476264,
        -0.064826645,
        0.1035554,
        0.023044914,
        0.07720165,
        -0.091542244,
        0.45808342,
        -0.05759049,
        0.039017595,
        -0.01578809,
        -0.087798834,
        0.032718167,
        -0.0078352615,
        -0.04595129,
        0.040975865,
        -0.00043430255,
        0.03269844,
        0.020269126,
        0.071435824,
        -0.033621114,
        0.0017961055,
        -0.041349147,
        -0.0014454057,
        -0.043340392,
        -0.00026739022,
        0.00006232308,
        -0.054089427,
        -0.057589754,
        -0.03377855,
        0.03559058,
        -0.02730587,
        0.05190788,
        0.017863173,
        -0.050276738,
        -0.065750845,
        0.10117645,
        -0.058037087,
        -0.08652497,
        0.01654462,
        -0.039831355,
        -0.08077718,
        -0.004733208,
        -0.060278844,
        0.07377389,
        -0.04541848,
        0.010558353,
        -0.11435882,
        0.035378337,
        0.050403196,
        0.013633781,
        0.057922848,
        -0.047061224,
        -0.09300169,
        -0.061433867,
        0.050096877,
        -0.041795444,
        0.023038669,
        -0.06598024,
        0.09703482,
        -0.04799871,
        -0.045372967,
        -0.010242292,
        0.0037029472,
        -0.08191514,
        0.008789093,
        -0.028667938,
        0.065494485,
        0.007818107,
        0.03273045,
        -0.036857612,
        -0.0058102217,
        -0.032093514,
        0.0043980847,
        -0.03181948,
        -0.017844534,
        -0.03055039,
        0.0033256179,
        -0.061082654,
        -0.044692542,
        -0.029833518,
        -0.012048393,
        0.0053293747,
        -0.09173112,
        -0.04654764,
        0.03879226,
        0.04326413,
        -0.02672302,
        0.03908756,
        -0.6163778,
        -0.03958774,
        -0.026941083,
        -0.042221412,
        -0.14455579,
        0.041197225,
        0.03861385,
        -0.026768295,
        0.11270465,
        -0.02228305,
        0.06662755,
        -0.0050237356,
        -0.032705493,
        -0.015536662,
        0.033559043,
        0.07644662,
        0.050653398
      ]
    },
    {
      "id": "worldsfair-speaker-301",
      "conference": "worldsfair",
      "name": "Todd Fisher",
      "role": "Head of Software Engineering",
      "company": "Philo Ventures",
      "twitter": "",
      "talks": [
        {
          "title": "While my guitar gently speaks",
          "description": "Do you ever wonder What the next evolution of live performances will look like? I do all the time. Come experience what happens when you combine live guitar playing with DSP as well as TTS and other models, all running locally. Prepare to be entertained and get familiar with new possibilities that modern AI opens up in the audio and digital signal processing space while you enjoy a live performance on top of an informative slide presentation. Walk away from this talk inspired to help build the next evolution of tools for musicians and live performances. We will touch on how to build with tools such as classic DSP, JUCE, on device TTS, CoreML, WhisperX, CoreMIDI and more! You might even get a chance to have a conversation with a guitar!",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.013256958,
        0.04934718,
        0.001957054,
        0.02382582,
        0.06795616,
        0.031100227,
        -0.10560861,
        0.06369713,
        0.06260619,
        -0.21367106,
        -0.051297117,
        -0.026932223,
        -0.016873688,
        -0.037417654,
        -0.096142925,
        0.0041975183,
        0.0866337,
        -0.06624211,
        -0.023452746,
        -0.1028686,
        -0.03772704,
        -0.114906296,
        -0.061848722,
        0.12652539,
        -0.09377238,
        0.117844746,
        0.016882725,
        0.082947746,
        -0.086447164,
        0.4377111,
        -0.09855847,
        0.011066574,
        0.0151730385,
        0.02198593,
        0.052815735,
        0.04266084,
        -0.051769998,
        0.06793393,
        -0.01161097,
        0.0028212855,
        0.015270749,
        -0.016526164,
        -0.020453218,
        0.08148949,
        -0.00090134493,
        -0.038881544,
        -0.053765927,
        0.020430742,
        0.010134475,
        0.00059860485,
        0.01765621,
        -0.04427056,
        0.06863421,
        -0.05259008,
        -0.0020301866,
        0.03209679,
        0.036130965,
        -0.0915535,
        0.09238966,
        -0.04189032,
        -0.10644926,
        0.00046993323,
        -0.025296744,
        0.004478805,
        0.027610773,
        -0.056342624,
        0.010295046,
        0.008898613,
        -0.07940352,
        -0.046446893,
        -0.00041246405,
        0.031816028,
        -0.08221152,
        0.023114972,
        -0.05092938,
        -0.09074362,
        -0.05144715,
        0.062889494,
        -0.006649405,
        0.011152903,
        0.07974694,
        0.008968802,
        -0.096846536,
        -0.046081584,
        0.11242539,
        -0.04783603,
        -0.025194183,
        0.03568623,
        -0.0017443539,
        0.09127292,
        -0.002115991,
        0.005238113,
        -0.052414406,
        0.00051706436,
        -0.008357096,
        0.031108396,
        -0.04052367,
        0.053666912,
        -0.0058329385,
        -0.05998412,
        -0.026483057,
        -0.009895671,
        -0.016976392,
        -0.025838912,
        -0.034959238,
        -0.07856071,
        0.0063469536,
        0.011501001,
        0.03826726,
        -0.003910211,
        0.015422698,
        -0.6170373,
        0.08671716,
        -0.05262105,
        -0.044067737,
        -0.15528916,
        0.030969683,
        -0.061292414,
        0.05684662,
        0.104428746,
        -0.029219314,
        0.012207176,
        -0.016965006,
        -0.013766288,
        0.011125949,
        0.070236444,
        0.0017101251,
        -0.028720517
      ]
    },
    {
      "id": "worldsfair-speaker-302",
      "conference": "worldsfair",
      "name": "Jan Curn",
      "role": "Founder & CEO",
      "company": "Apify",
      "twitter": "https://x.com/jancurn",
      "talks": [
        {
          "title": "x402 isn’t good (yet)",
          "description": "While everyone understands that agents will get more done with a budget, no one knows which protocol will win agentic payment standard wars: x402, MPP, Skyfire, or another? So far, x402 is the most mature protocol with the largest transaction volume, but even its new \"upto\" payment scheme doesn’t support true usage-based pricing, as it gives agents a chance to consume resources and then skip out on the bill. I’ll walk you through our experience (and pains) implementing agentic payments for a marketplace of 30K+ web Actors, and how we made it work even with the current specs.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.06953269,
        0.07050915,
        0.0047452445,
        -0.03957529,
        -0.020251136,
        0.004184731,
        -0.08153164,
        -0.03768326,
        0.02953319,
        -0.26599225,
        -0.081953265,
        -0.015011626,
        0.03267696,
        -0.077526115,
        -0.114675775,
        -0.0019670704,
        0.083997354,
        -0.08529146,
        -0.10592969,
        -0.065521196,
        -0.08060972,
        -0.10799637,
        -0.021523643,
        0.015585497,
        0.06251795,
        0.105167694,
        -0.020744616,
        0.043371763,
        -0.11152412,
        0.4350423,
        -0.09369561,
        -0.017963445,
        0.013671781,
        -0.027396495,
        0.015044498,
        -0.020053977,
        -0.09194324,
        -0.039351337,
        0.05923874,
        0.023363197,
        -0.024335258,
        0.05037548,
        -0.06710183,
        0.03994633,
        -0.05787172,
        0.03323442,
        -0.008392921,
        -0.0019517523,
        -0.017547933,
        0.015799822,
        -0.087084755,
        0.02179127,
        0.0047094333,
        0.0012683603,
        0.05252905,
        -0.022555985,
        0.06524094,
        -0.04375521,
        0.11548519,
        -0.054886095,
        -0.17102607,
        -0.0151011385,
        -0.027653411,
        0.05997622,
        -0.063640036,
        -0.08749165,
        0.06091091,
        -0.070179515,
        0.02378704,
        -0.046446513,
        0.003576914,
        0.08556585,
        -0.028305924,
        0.0004034422,
        -0.018120145,
        -0.07668543,
        -0.010998316,
        0.080763824,
        -0.022799892,
        0.007661636,
        -0.0032544634,
        0.07568656,
        -0.023854941,
        0.024729695,
        0.014688,
        -0.024424152,
        -0.05422963,
        0.008645913,
        -0.033215955,
        0.063224114,
        -0.052805506,
        -0.049287003,
        -0.050806705,
        0.025571419,
        -0.051553037,
        0.029791601,
        -0.119467854,
        0.015183679,
        -0.04236609,
        0.07583528,
        0.02731197,
        -0.013544397,
        0.017256685,
        -0.043863297,
        0.010537422,
        -0.01013146,
        0.06032809,
        -0.0028194608,
        0.029630508,
        0.013171181,
        -0.0253631,
        -0.5818482,
        -0.015421194,
        0.020066518,
        0.005756833,
        -0.13805847,
        0.048801027,
        0.02291274,
        0.040011317,
        0.12669227,
        -0.07085996,
        0.03457279,
        -0.031779822,
        0.07471389,
        -0.042601082,
        0.04286774,
        0.008740013,
        0.090895146
      ]
    },
    {
      "id": "worldsfair-speaker-303",
      "conference": "worldsfair",
      "name": "Lydia Hallie",
      "role": "Claude Code team",
      "company": "Anthropic",
      "twitter": "https://x.com/lydiahallie",
      "talks": [
        {
          "title": "Claude Cowork in Finance",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.005372611,
        -0.005346198,
        0.0367427,
        0.015627904,
        0.031963054,
        -0.04212055,
        -0.067544736,
        0.026343964,
        0.05771709,
        -0.19057548,
        -0.041652806,
        0.004882811,
        0.052688636,
        -0.0030861453,
        -0.07782248,
        0.05487778,
        0.08226667,
        -0.055567518,
        -0.007880358,
        -0.12093748,
        -0.06945718,
        -0.0013743655,
        0.052799657,
        0.0569518,
        -0.08253202,
        0.12643412,
        -0.017375194,
        0.021585546,
        -0.13063778,
        0.44998324,
        -0.14558731,
        0.024341408,
        0.047958132,
        0.010964648,
        0.065873966,
        -0.04789598,
        -0.038071185,
        -0.010578389,
        0.06769872,
        0.08898162,
        0.03554683,
        0.051970042,
        -0.025114415,
        0.032134492,
        -0.031496394,
        0.041649748,
        -0.12721899,
        0.004604051,
        -0.053246856,
        0.0038401901,
        -0.037352365,
        -0.030658832,
        0.011917693,
        -0.012280278,
        0.07313218,
        0.045385137,
        -0.036640283,
        -0.10616727,
        0.11177376,
        -0.06596296,
        -0.1009368,
        -0.020318283,
        -0.019041376,
        0.015250544,
        0.053617403,
        -0.030591644,
        0.02586639,
        -0.048369784,
        0.017790195,
        -0.05297065,
        -0.013688984,
        0.0208674,
        -0.050246082,
        0.021526393,
        -0.058983058,
        -0.09414265,
        0.008509406,
        0.078890584,
        -0.03371432,
        -0.054310963,
        0.03018108,
        0.08234867,
        -0.120685965,
        -0.0417315,
        0.020615563,
        -0.004510923,
        -0.045784883,
        0.059724838,
        0.025134703,
        0.03762997,
        0.017623179,
        -0.042795897,
        -0.010473609,
        -0.019719586,
        -0.12970778,
        0.012873493,
        -0.006385497,
        0.0148961535,
        0.021438127,
        0.0013176732,
        0.010984837,
        -0.10530797,
        0.041725192,
        -0.023050098,
        -0.060753684,
        0.005351936,
        -0.018541222,
        0.052434854,
        0.03819007,
        0.016229084,
        -0.07301585,
        -0.5860809,
        0.045116037,
        0.05874758,
        -0.09861937,
        -0.16454013,
        -0.040625013,
        -0.043425735,
        0.018965436,
        0.11307092,
        -0.02091979,
        0.08032827,
        -0.04983876,
        0.0074782525,
        -0.027259575,
        0.04429256,
        0.06275441,
        -0.029070914
      ]
    },
    {
      "id": "worldsfair-speaker-304",
      "conference": "worldsfair",
      "name": "Denys Linkov",
      "role": "Head of ML",
      "company": "Wisedocs",
      "twitter": "",
      "talks": [
        {
          "title": "Benchmarking Coding Agents on New vs Legacy Code bases",
          "description": "You have an old code base with 100,000s of lines of code, should you let an AI Agent refactor or do you wait until you have a cleaner setup? Last year we refactored a number of code bases and ran evaluations on how well different models, harnesses and rule sets affected multiple versions of the code base. This talk will feature specific code examples as well as a broader set of evals.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03743788,
        0.048873726,
        0.029807402,
        0.02705263,
        -0.03349434,
        0.0058022793,
        -0.0781906,
        0.008181276,
        0.084150776,
        -0.21587586,
        -0.056769814,
        -0.017724149,
        0.046767384,
        0.06495858,
        -0.05370288,
        -0.027960528,
        0.1630731,
        -0.031569373,
        -0.09942032,
        -0.0618473,
        -0.03176374,
        -0.02069598,
        0.016617013,
        0.0497346,
        -0.06251381,
        0.0867005,
        -0.039515097,
        0.06023594,
        -0.10259392,
        0.46551642,
        -0.106756546,
        0.03986864,
        0.018196402,
        -0.04188686,
        -0.014585628,
        -0.08936838,
        -0.0667256,
        0.011011464,
        -0.0023690695,
        0.0505778,
        -0.016149256,
        0.040267866,
        -0.022324076,
        0.036678787,
        -0.012037578,
        -0.024439486,
        -0.055605095,
        0.0704363,
        -0.0037002533,
        -0.069528334,
        -0.06707586,
        0.039592937,
        0.048006535,
        -0.008450018,
        0.027588308,
        -0.023508802,
        0.03267051,
        -0.09590352,
        0.07692817,
        -0.035351396,
        -0.079450704,
        -0.015437814,
        0.027721658,
        0.017041944,
        -0.0076810275,
        -0.08150771,
        0.028669449,
        -0.049553424,
        -0.009779634,
        -0.043563843,
        -0.022648932,
        0.0326885,
        -0.09522675,
        0.050671525,
        -0.09933142,
        -0.08793311,
        -0.044486467,
        0.10303772,
        -0.03519177,
        0.09649998,
        -0.02630553,
        0.12385917,
        -0.053460706,
        0.0060465857,
        0.08860818,
        0.084576875,
        0.0017969114,
        0.058105245,
        -0.04782912,
        0.07345118,
        0.051741257,
        -0.06268039,
        -0.09329115,
        -0.009837904,
        -0.10084519,
        0.06853286,
        -0.076588005,
        0.011671213,
        -0.0006934444,
        0.0060368376,
        0.019753402,
        -0.01878573,
        0.0312004,
        0.025296358,
        0.052837618,
        -0.029805701,
        -0.038854703,
        -0.01408434,
        0.07082646,
        -0.014643578,
        0.021802302,
        -0.58325875,
        0.0022557576,
        0.0045420392,
        -0.05943862,
        -0.15456408,
        0.03421929,
        0.024983535,
        -0.0012187305,
        0.068871655,
        -0.020433292,
        0.09447985,
        -0.015331425,
        -0.021365954,
        0.020066248,
        0.021062149,
        0.045372553,
        0.015446065
      ]
    },
    {
      "id": "worldsfair-speaker-305",
      "conference": "worldsfair",
      "name": "Mike Phipps",
      "role": "Lead AI Engineer",
      "company": "Gates Foundation",
      "twitter": "",
      "talks": [
        {
          "title": "Your Moat Is Your Data Model",
          "description": "Every enterprise AI team faces the same strategic question: where in the stack should a small team focus its effort? Models, frontends, and agent frameworks evolve rapidly and are increasingly commoditized. But regardless of how these layers mature, AI in enterprise settings remains bottlenecked by the same underlying problem: structured data is siloed across systems of record with domain-specific schemas, and the unstructured data needed to contextualize it sits in entirely separate systems, with its own systematic complexities. The durable work is cleaning, curating, and semantically modeling this data in an AI-first manner so that any client — chat, workflow, or otherwise — can query across it. That's the moat. At the Gates Foundation, my team built and deployed our foundation-wide knowledge graph on Neo4j that unifies structured and unstructured data behind a single MCP server. The graph itself is modeled for agentic consumption: natural hierarchies are projected as traversable paths rather than flattened tables, and unstructured documents are semantically chunked, tagged, and mapped to structured entities at ingestion time using AI-driven ETL. The result is a semantic layer where an agent can express a complex cross-system question as a concise graph query and receive an accurate answer. This talk is an architectural walkthrough covering the end-to-end pipeline: AI-based extraction and semantic chunking of unstructured documents, the agent-first data modeling decisions, design considerations for our MCP server, and how we handle graph-based retrieval evals. We'll walk through real query sessions showing Claude interacting with the graph through both chat and workflow integrations. The intended takeaway is a practical framework for where a small enterprise team's investment compounds — and why that investment is the data model, not the layers above it.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.031607546,
        0.060622055,
        0.063705355,
        -0.014347198,
        -0.0108109135,
        0.004558618,
        -0.04839678,
        -0.01640206,
        0.016941914,
        -0.23584439,
        -0.026841993,
        -0.08725551,
        -0.015430505,
        -0.057662833,
        -0.045600973,
        -0.0051881066,
        0.06672728,
        -0.04815896,
        -0.087650836,
        -0.041490175,
        -0.015942797,
        -0.044514123,
        0.0028869808,
        0.072646916,
        -0.05619516,
        0.10751882,
        0.043828405,
        0.084903404,
        -0.10826645,
        0.45266843,
        -0.09560962,
        0.06765503,
        -0.068193175,
        -0.0012883737,
        0.013985318,
        -0.015038261,
        -0.06732993,
        -0.043730777,
        -0.010244729,
        0.009523954,
        -0.018641347,
        0.038072836,
        -0.075579815,
        0.0014588209,
        -0.012670471,
        -0.016933672,
        -0.05913651,
        -0.004919673,
        -0.0058280057,
        0.00010133707,
        -0.11609479,
        -0.019078897,
        0.050579898,
        0.020067556,
        0.06028068,
        0.06554005,
        -0.01563246,
        -0.051157735,
        0.11203257,
        -0.08574608,
        -0.11728746,
        0.02101248,
        -0.012674453,
        -0.02848741,
        0.008481649,
        -0.029044317,
        0.01579241,
        0.00418623,
        -0.037924457,
        -0.047348965,
        0.009065773,
        0.0189852,
        -0.087657034,
        0.05532867,
        -0.058639526,
        -0.088800974,
        -0.0017151652,
        0.040076107,
        -0.05848478,
        0.035478625,
        0.047995288,
        0.040278934,
        -0.013285792,
        -0.08624826,
        -0.0149285365,
        -0.04712989,
        -0.09138436,
        0.03453998,
        0.056947228,
        0.04888862,
        -0.060578845,
        -0.010819488,
        -0.0469756,
        0.063804045,
        -0.091552086,
        0.009893086,
        -0.02656483,
        0.01527333,
        -0.059408266,
        0.03976144,
        -0.006741328,
        -0.07643839,
        0.04890623,
        -0.024933085,
        -0.005312111,
        -0.12900242,
        0.027239384,
        -0.020135026,
        0.09814655,
        -0.023849066,
        -0.081718944,
        -0.60679644,
        -0.00551921,
        0.017766282,
        0.050790668,
        -0.14994787,
        0.03337015,
        -0.02571738,
        0.047476545,
        0.08933998,
        0.009038045,
        0.057418205,
        0.00017261949,
        -0.04480782,
        0.008842349,
        0.035329424,
        0.092870556,
        0.051298782
      ]
    },
    {
      "id": "worldsfair-speaker-306",
      "conference": "worldsfair",
      "name": "Arman Vaziri",
      "role": "GTM",
      "company": "Ramp",
      "twitter": "",
      "talks": [
        {
          "title": "Ramp: AI in GTM",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0029952982,
        0.08300972,
        0.05797838,
        -0.02806859,
        0.040608987,
        0.030249529,
        -0.06850457,
        0.070539646,
        0.08525292,
        -0.2384262,
        0.022858752,
        -0.031110724,
        0.009103979,
        -0.087961115,
        -0.042820327,
        0.006239295,
        0.08519878,
        -0.058106314,
        -0.064831115,
        -0.09089664,
        -0.06660939,
        -0.064040646,
        0.05550777,
        0.04623963,
        -0.03132706,
        0.04119951,
        0.014015075,
        0.031972665,
        -0.06544425,
        0.47221398,
        -0.067336924,
        -0.0011741117,
        -0.001622217,
        0.0008633458,
        0.010399697,
        -0.09687981,
        -0.015123257,
        -0.0020716512,
        0.026300369,
        0.09543831,
        -0.019584317,
        0.052160304,
        -0.00804441,
        -0.011396211,
        -0.045596868,
        0.023371598,
        -0.08798866,
        0.004193807,
        -0.034220736,
        -0.057500374,
        -0.05990473,
        0.026713409,
        0.03291125,
        -0.020409534,
        0.049069244,
        0.012408427,
        -0.035621405,
        -0.07353257,
        0.08227493,
        -0.03394122,
        -0.12568054,
        0.022192258,
        0.013863479,
        0.0146075,
        -0.02600947,
        -0.043522898,
        0.042248297,
        -0.023953991,
        -0.087941095,
        -0.05366108,
        0.0032485507,
        0.07527817,
        -0.11865089,
        0.05587132,
        -0.07869835,
        -0.077516094,
        0.026978344,
        0.05541358,
        0.029024478,
        -0.02981771,
        0.07754211,
        0.06602466,
        -0.08862608,
        -0.009974788,
        0.07650709,
        -0.045117266,
        -0.015438668,
        0.025626885,
        -0.014162642,
        0.0348514,
        0.04344593,
        -0.0018510831,
        -0.0030278463,
        -0.047246244,
        -0.046130113,
        0.015328615,
        -0.07696585,
        0.05388556,
        -0.016283248,
        0.03744127,
        0.016305119,
        -0.058960684,
        0.09498174,
        0.025312707,
        -0.005259908,
        -0.025640516,
        0.029524814,
        -0.017261354,
        -0.0053809034,
        -0.003501539,
        -0.095253125,
        -0.60895634,
        -0.0044644764,
        0.027556583,
        -0.0054187994,
        -0.13727602,
        -0.027390353,
        -0.014374352,
        0.08013872,
        0.061105996,
        -0.04882603,
        -0.0095140925,
        -0.05965202,
        -0.02334865,
        0.024838705,
        0.021605313,
        0.111890495,
        -0.060815793
      ]
    },
    {
      "id": "worldsfair-speaker-307",
      "conference": "worldsfair",
      "name": "Vivek Raju Muppalla",
      "role": "VP of AI Engineering",
      "company": "Hippocratic AI",
      "twitter": "https://x.com/vim1up",
      "talks": [
        {
          "title": "200 Million Patient Interactions Later: What the Generic Voice Stack Misses",
          "description": "A healthcare voice agent can be right on the benchmark and still fail in production. Real patients hesitate, interrupt, misremember medications, code-switch mid-sentence, and disclose risk indirectly. After 200M+ patient-agent interactions, the lesson is clear: in clinical voice AI, interaction is a safety variable. This talk breaks down what Hippocratic AI had to rebuild beyond the generic voice stack: not just ASR, VAD, an LLM, TTS, and turn-taking heuristics, but a real-time safety system that treats silence, clarification, escalation, multilingual continuity, and medication-specific recognition as first-class engineering problems. We’ll walk through the production architecture behind Hippocratic AI’s voice agents: a 30+ model supervisor constellation, including the 4.1T-parameter AI Front Door system, designed to catch failures a single primary model misses. The talk covers how specialized models monitor medication identification, overdose risk, labs and vitals, escalation criteria, workflow confirmation, and other clinical safety surfaces while the patient conversation is still happening. We’ll focus on four production lessons: Benchmarks are not enough; Interaction signals become training data; One LLM is not a safety architecture; Voice infrastructure has clinical failure modes.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03586797,
        0.0761395,
        0.019154023,
        0.036022104,
        0.020764709,
        0.037987538,
        -0.12545191,
        0.012496222,
        0.068653435,
        -0.20006418,
        0.016866611,
        -0.052236408,
        0.061178092,
        -0.13158464,
        -0.07979007,
        0.003746704,
        0.050941415,
        -0.06688933,
        -0.07135106,
        -0.09689937,
        -0.015858835,
        -0.06744766,
        -0.024369072,
        0.066843025,
        -0.07150517,
        0.033938263,
        0.066296324,
        0.05038346,
        -0.09280768,
        0.42663813,
        -0.12645221,
        0.04198126,
        -0.00008666335,
        -0.0020556792,
        0.08689652,
        0.0020662344,
        -0.016436245,
        0.01677985,
        -0.0040531284,
        0.07576784,
        0.04064642,
        0.074638166,
        -0.04727065,
        -0.0053569,
        0.035078045,
        -0.07783199,
        -0.059129532,
        0.06580735,
        -0.020576758,
        -0.0076616355,
        -0.08528735,
        -0.02710386,
        0.027019892,
        -0.03109685,
        0.028719258,
        0.03025238,
        0.011214144,
        -0.059667118,
        0.046738155,
        -0.033118766,
        -0.13569126,
        0.014449853,
        0.07217868,
        -0.037954424,
        -0.032091167,
        -0.124138646,
        0.03452464,
        0.02348443,
        -0.08380284,
        0.0064627444,
        -0.048096087,
        0.032891486,
        -0.044138446,
        -0.0088282125,
        -0.015627034,
        -0.12324566,
        -0.021674875,
        0.074825056,
        -0.006481238,
        -0.0170489,
        -0.01495917,
        0.06553942,
        -0.08042595,
        -0.025525521,
        0.03805286,
        -0.06656627,
        -0.04342823,
        0.037143897,
        -0.060430467,
        0.113946296,
        0.037783287,
        -0.020070763,
        -0.009187704,
        0.0045783455,
        -0.025439382,
        0.032133706,
        -0.04887455,
        -0.007555405,
        -0.025429325,
        0.06615024,
        0.0446954,
        -0.012919033,
        0.071048945,
        -0.02305278,
        0.025336785,
        -0.004260505,
        0.0073122317,
        -0.014694381,
        0.04488858,
        -0.013734361,
        -0.11249952,
        -0.62323755,
        0.04281577,
        0.02056487,
        0.00037112267,
        -0.15179253,
        0.05545495,
        -0.00757596,
        0.05456781,
        0.05760899,
        -0.04834337,
        0.03695922,
        -0.007883954,
        0.0011513998,
        -0.043409586,
        0.04339396,
        0.085653655,
        -0.009872642
      ]
    },
    {
      "id": "worldsfair-speaker-308",
      "conference": "worldsfair",
      "name": "Rita Zhang",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Vertical Mobility: Building an AI Inference Platform That Scales from MVP to Trillion-Parameter Workloads",
          "description": "The future of AI inference is not one-size-fits-all. This talk explores a multi-tiered architecture that supports the full AI lifecycle, from rapid, pay-per-token experimentation to dedicated, SLO-bound production and extreme-scale, self-managed deployments. Learn about lessons learned from CoreWeave’s inference stack as performance, cost, and control requirements evolve.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.08742635,
        0.10132034,
        0.037639827,
        0.04140554,
        0.064089745,
        -0.07311889,
        0.010127131,
        -0.03339356,
        0.0624151,
        -0.2691385,
        0.0010538427,
        -0.017910006,
        -0.024009047,
        0.00018262463,
        -0.036975615,
        0.05269874,
        0.0585131,
        -0.051544487,
        -0.06324389,
        -0.012106047,
        -0.07171721,
        -0.06452995,
        0.0027670558,
        0.062009692,
        -0.03651025,
        0.11085547,
        0.069360666,
        0.009219918,
        -0.10125304,
        0.42917463,
        -0.07837175,
        -0.02231041,
        -0.023599517,
        -0.04540303,
        0.04075223,
        -0.04902581,
        0.0027839267,
        -0.02405501,
        0.06250681,
        0.041021883,
        0.045033354,
        0.04941814,
        -0.06676506,
        0.015576571,
        -0.085785136,
        -0.020531384,
        -0.10158046,
        0.01772958,
        -0.023675332,
        0.025933947,
        -0.060943212,
        0.035623554,
        0.046316434,
        0.011598162,
        -0.020706208,
        0.015241862,
        -0.08951319,
        -0.09155543,
        0.12887265,
        -0.0048261415,
        -0.11919606,
        0.04953162,
        -0.0037184926,
        0.009648027,
        0.012110635,
        -0.03538201,
        -0.021675581,
        -0.06884429,
        -0.0064385016,
        -0.06042361,
        0.009079587,
        0.026317814,
        -0.033076495,
        0.010673055,
        -0.042671937,
        -0.15656792,
        -0.019355867,
        0.061806835,
        -0.002088863,
        0.010212648,
        -0.027862484,
        0.06962552,
        -0.061728567,
        -0.0028695674,
        0.040877122,
        0.027736884,
        -0.08833072,
        0.06744074,
        -0.049337994,
        0.10789468,
        0.027333843,
        -0.013753624,
        -0.007903203,
        0.00038561475,
        -0.08933349,
        -0.0098757455,
        -0.022808691,
        0.04844313,
        0.0074224053,
        0.070269,
        -0.0022772837,
        0.013909419,
        -0.021347145,
        0.012901186,
        -0.0057759834,
        -0.045346923,
        -0.010089192,
        -0.042106785,
        0.021926515,
        -0.018181851,
        -0.092877306,
        -0.6218352,
        0.026645424,
        0.04710305,
        -0.06830728,
        -0.13973884,
        0.00338735,
        -0.043136753,
        0.030175155,
        0.05136493,
        -0.04633494,
        0.071688555,
        -0.045577083,
        -0.014760808,
        0.013653445,
        0.011778901,
        0.05199457,
        0.017623646
      ]
    },
    {
      "id": "worldsfair-speaker-309",
      "conference": "worldsfair",
      "name": "Sitanshu Gupta",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Vertical Mobility: Building an AI Inference Platform That Scales from MVP to Trillion-Parameter Workloads",
          "description": "The future of AI inference is not one-size-fits-all. This talk explores a multi-tiered architecture that supports the full AI lifecycle, from rapid, pay-per-token experimentation to dedicated, SLO-bound production and extreme-scale, self-managed deployments. Learn about lessons learned from CoreWeave’s inference stack as performance, cost, and control requirements evolve.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.052691452,
        0.10355867,
        -0.008616694,
        0.04335049,
        0.06749563,
        -0.03749222,
        -0.017110743,
        -0.055017993,
        0.0404755,
        -0.25725123,
        -0.043877073,
        0.0093350345,
        -0.009782179,
        -0.051069345,
        -0.0044735717,
        0.010865688,
        0.05699104,
        -0.06852001,
        -0.045631785,
        0.006490427,
        -0.032901134,
        -0.089785956,
        0.0031982614,
        0.06747778,
        -0.000701461,
        0.0794896,
        0.025545523,
        0.025131814,
        -0.1026496,
        0.42966563,
        -0.06295159,
        -0.0131274965,
        -0.012902305,
        -0.054605324,
        0.0381172,
        -0.030228283,
        -0.06827627,
        0.024101831,
        0.06684056,
        0.05306465,
        0.04218791,
        0.06971447,
        -0.07976355,
        0.008178165,
        -0.043879367,
        -0.06370102,
        -0.10769496,
        -0.006430891,
        -0.015031986,
        -0.005335032,
        -0.091189206,
        0.010854741,
        0.04947415,
        0.025307627,
        -0.02774875,
        0.031426433,
        -0.07762538,
        -0.067587234,
        0.09768611,
        -0.04123924,
        -0.14950427,
        -0.024457268,
        0.031474493,
        0.018600594,
        0.022660203,
        -0.06342953,
        0.033615947,
        -0.07026306,
        -0.0038665184,
        -0.05786204,
        -0.014477967,
        0.029124163,
        -0.07170826,
        0.017216152,
        0.013436856,
        -0.14457384,
        -0.011751069,
        0.05165831,
        0.000084778665,
        -0.022962622,
        -0.038849287,
        0.061600327,
        -0.006288242,
        0.0028187071,
        0.045856442,
        -0.02794116,
        -0.048747238,
        0.040896237,
        -0.00785827,
        0.10506536,
        0.04611846,
        0.0006457236,
        -0.023727173,
        0.0006506534,
        -0.011787278,
        -0.0065745027,
        -0.036590107,
        0.0029359132,
        -0.0075811627,
        0.05149766,
        0.05561613,
        -0.013354513,
        -0.015428933,
        0.031945202,
        0.015506397,
        -0.03205597,
        0.07499909,
        -0.020615378,
        0.013660718,
        0.042616,
        -0.04895506,
        -0.6477719,
        0.026465246,
        0.06275085,
        -0.08695496,
        -0.14776094,
        0.0022133838,
        -0.051753163,
        0.025448678,
        0.049547415,
        -0.058397,
        0.049083296,
        -0.02727952,
        -0.056935564,
        0.018028522,
        -0.0011402131,
        0.049571726,
        -0.009244535
      ]
    },
    {
      "id": "worldsfair-speaker-310",
      "conference": "worldsfair",
      "name": "Lin Qiao",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Stop Model Shopping: Why Ownership Beats Choice in the Agent Stack",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.028109243,
        0.08174729,
        -0.017232224,
        0.035993304,
        0.013298516,
        -0.05583251,
        -0.057477232,
        0.017483039,
        0.021846296,
        -0.21774913,
        0.015516162,
        0.01610075,
        -0.013960908,
        0.038858727,
        -0.021000916,
        0.035364196,
        0.08600888,
        -0.08019102,
        -0.116459206,
        -0.08283088,
        -0.07694549,
        -0.11919738,
        0.0051996163,
        0.003601406,
        -0.06452005,
        0.0400626,
        0.019726234,
        0.037192076,
        -0.09013192,
        0.47041738,
        -0.10166442,
        0.028004631,
        -0.01860459,
        -0.056163866,
        0.054614995,
        -0.03921305,
        0.026448015,
        -0.017675774,
        0.052928682,
        0.051053718,
        0.06913189,
        0.034017213,
        -0.051928658,
        0.057929598,
        -0.07858297,
        -0.019517772,
        -0.03164451,
        -0.00093783735,
        -0.051032934,
        -0.033988114,
        -0.028752258,
        -0.043697324,
        0.013437795,
        -0.07924923,
        0.023610948,
        -0.016795086,
        -0.023681147,
        -0.004637253,
        0.09164891,
        0.03910033,
        -0.14767519,
        0.0068655033,
        -0.071647115,
        0.00020287537,
        -0.0059706583,
        -0.03187958,
        0.061212115,
        -0.09523573,
        0.020147622,
        -0.067329176,
        0.03890706,
        -0.0059078205,
        -0.01595244,
        0.02090434,
        -0.114665166,
        -0.0759085,
        0.015451682,
        0.055797506,
        -0.09478204,
        -0.032605555,
        0.04217622,
        0.058462087,
        -0.11471371,
        -0.10916904,
        0.015932798,
        -0.017774137,
        -0.09106204,
        0.05810622,
        -0.053073708,
        0.065299585,
        0.03906138,
        0.07177847,
        -0.01527028,
        0.01991227,
        -0.072476014,
        -0.0382709,
        -0.07995819,
        0.057445638,
        -0.012349611,
        -0.004707456,
        0.010037151,
        0.006620607,
        0.0646305,
        -0.033149365,
        -0.020268336,
        -0.04597833,
        -0.03285406,
        -0.019938057,
        0.080185615,
        -0.04271205,
        -0.04330012,
        -0.5966738,
        -0.009601624,
        0.014151601,
        -0.051272158,
        -0.13500254,
        0.088471696,
        0.016305482,
        -0.02969489,
        0.05258071,
        -0.08863194,
        0.026612181,
        0.007965713,
        -0.0121748205,
        0.005744602,
        0.002338572,
        -0.012503252,
        -0.013839228
      ]
    },
    {
      "id": "worldsfair-speaker-311",
      "conference": "worldsfair",
      "name": "Giedrius Steimantas",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Missing Layer in Agentic AI",
          "description": "Reasoning is solved. Web access isn't. Most agents break the moment they leave the sandbox blocked, rate-limited, or staring at a CAPTCHA. Giedrius will show the three primitives every production agent needs: a browser, a fast search API, and a universal scraper and demo an agent built on top of them that actually works in the wild.",
          "day": "Day 4 — Session Day 3",
          "time": "12:05pm-12:25pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.02761264,
        0.060588606,
        0.0824453,
        0.015333998,
        0.011495166,
        -0.0021111008,
        -0.033362336,
        -0.03519294,
        0.041397616,
        -0.21427028,
        -0.029012341,
        0.0048451046,
        -0.03555733,
        -0.022328127,
        0.011167852,
        0.028791226,
        0.08810446,
        -0.09003199,
        -0.072748244,
        -0.06304042,
        -0.055704657,
        -0.10606571,
        0.015220041,
        0.04924759,
        0.020218758,
        0.06956519,
        -0.018371426,
        0.041860353,
        -0.113259286,
        0.46772078,
        -0.041461382,
        -0.026540799,
        -0.009569148,
        0.01384637,
        -0.012722469,
        -0.0665145,
        -0.061018635,
        -0.026016716,
        0.03767103,
        0.05919315,
        -0.026188748,
        0.08809002,
        0.03276082,
        -0.010228989,
        -0.001344015,
        -0.036189605,
        -0.07075199,
        0.0006511363,
        -0.03808535,
        -0.018960303,
        -0.08432737,
        0.078227244,
        0.036556087,
        0.0014025508,
        0.03518381,
        -0.0022456308,
        -0.0035929896,
        -0.09684924,
        0.0780434,
        -0.06651402,
        -0.11541924,
        -0.041286845,
        -0.016766885,
        0.047309633,
        0.024146061,
        -0.10247854,
        0.07836523,
        -0.12643418,
        0.047505785,
        -0.07965998,
        0.0000026858813,
        0.026007803,
        -0.07135183,
        0.07211935,
        -0.030848747,
        -0.07770155,
        -0.05048021,
        0.07427595,
        -0.040246762,
        0.017815001,
        -0.06689361,
        0.05723894,
        0.0017681329,
        0.0072011063,
        0.047066674,
        -0.017122013,
        -0.06199346,
        0.0029668245,
        0.0024933878,
        0.06907428,
        0.03873326,
        -0.025780365,
        -0.077196546,
        0.0072954823,
        -0.011711884,
        -0.0025029287,
        -0.047474775,
        -0.00675992,
        -0.0035920325,
        0.038625818,
        -0.04309535,
        0.022897186,
        0.018994218,
        0.022789598,
        0.03445449,
        0.0030232011,
        0.03298176,
        0.0029103002,
        0.055084538,
        -0.05701639,
        0.004370383,
        -0.63906896,
        0.0026126562,
        0.018314809,
        0.020719279,
        -0.1640124,
        -0.010906109,
        0.016306875,
        0.016038546,
        0.07102317,
        -0.014931504,
        0.08806296,
        -0.00007900136,
        0.019267105,
        -0.0033819887,
        0.03863095,
        0.03307468,
        0.026792068
      ]
    },
    {
      "id": "worldsfair-speaker-312",
      "conference": "worldsfair",
      "name": "Rodrigo Coelho",
      "role": "",
      "company": "Edge & Node",
      "twitter": "https://x.com/rodventures",
      "talks": [
        {
          "title": "Agent Spending Without Controls: The Missing Infrastructure Layer for AI Pa…",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.005300843,
        0.010794644,
        0.036079098,
        -0.0011879513,
        0.008187561,
        0.022177622,
        -0.034286,
        0.0020902322,
        0.039972365,
        -0.19429111,
        -0.016825872,
        -0.085072465,
        0.008331068,
        -0.00006094435,
        -0.07786063,
        0.014915197,
        0.060361028,
        -0.10975372,
        -0.027490765,
        -0.091001496,
        -0.09525414,
        -0.08555568,
        0.028664794,
        -0.018503277,
        0.0070773354,
        0.055035934,
        -0.011024982,
        0.043839436,
        -0.09989788,
        0.53115445,
        -0.0132343145,
        -0.014296861,
        0.017945528,
        0.018303081,
        0.063734725,
        -0.050457425,
        -0.0015080222,
        0.0051650675,
        -0.010814109,
        0.058008395,
        -0.0075092707,
        0.083312064,
        -0.05689795,
        0.04774154,
        -0.053114526,
        0.0019615048,
        -0.04419803,
        -0.0024475253,
        -0.02786808,
        -0.012001444,
        -0.10034463,
        0.019653337,
        0.0148522835,
        -0.062284295,
        0.08088259,
        0.009739198,
        -0.036581293,
        -0.10723408,
        0.06111546,
        -0.007697127,
        -0.100735664,
        -0.051953863,
        -0.026449762,
        0.028374994,
        0.04203579,
        -0.032509107,
        0.05084301,
        -0.064077154,
        0.038847778,
        -0.08719487,
        0.02170349,
        -0.017528852,
        0.028047983,
        0.022319833,
        0.0041985926,
        -0.083729066,
        -0.03537262,
        0.034266554,
        -0.031026483,
        0.024768993,
        0.038732585,
        -0.0057426663,
        -0.029692879,
        -0.02182222,
        0.031077689,
        -0.0037303127,
        -0.04018297,
        -0.0045015193,
        -0.026109518,
        0.049082786,
        0.056981374,
        -0.014993309,
        0.008732509,
        -0.05948572,
        -0.04896545,
        0.09845684,
        -0.061615787,
        0.06828552,
        0.025008094,
        -0.007868256,
        -0.043727424,
        -0.04163172,
        0.049614813,
        -0.042910863,
        -0.011820402,
        -0.09418369,
        -0.0078050126,
        -0.013550311,
        0.057579987,
        -0.04812721,
        -0.06497882,
        -0.61778814,
        -0.038410734,
        -0.07937114,
        0.022105394,
        -0.15667288,
        -0.052402563,
        0.019354362,
        0.009979282,
        0.049646463,
        -0.03748981,
        0.053023864,
        -0.02749281,
        0.037042934,
        -0.0028818473,
        0.0044855736,
        0.017326208,
        -0.0016788273
      ]
    },
    {
      "id": "worldsfair-speaker-313",
      "conference": "worldsfair",
      "name": "Pranav Maheshwari",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Spending Without Controls: The Missing Infrastructure Layer for AI Pa…",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.020603327,
        0.062908344,
        0.041536182,
        0.03226997,
        0.005223243,
        -0.07420439,
        -0.048829973,
        0.014366558,
        0.10019059,
        -0.25554743,
        -0.08287286,
        -0.0073083113,
        0.0004780615,
        -0.06028665,
        -0.025027864,
        -0.027630987,
        0.04109451,
        -0.0015836017,
        -0.015271457,
        -0.03993932,
        -0.040146828,
        -0.055277485,
        -0.02690267,
        0.021885905,
        -0.03213676,
        0.0038976325,
        -0.024748279,
        0.010856674,
        -0.10333933,
        0.5042797,
        -0.0662697,
        0.034230266,
        -0.020265903,
        -0.03678579,
        0.0012949879,
        -0.03802484,
        -0.043830957,
        0.0062347646,
        0.024544908,
        0.04393196,
        -0.030633759,
        0.028907763,
        -0.037031025,
        -0.0048965826,
        -0.026038075,
        0.015552597,
        -0.007359332,
        0.008420805,
        0.05087707,
        -0.026730912,
        -0.056446526,
        -0.023833169,
        0.036743768,
        -0.044121973,
        0.04479729,
        0.038367294,
        0.009316708,
        -0.07738949,
        0.11805738,
        -0.023727618,
        -0.087926775,
        -0.0407585,
        -0.018755287,
        -0.053353924,
        -0.013876147,
        -0.060665,
        0.13147597,
        -0.04745532,
        0.05264255,
        0.029586103,
        -0.015459083,
        -0.047762107,
        -0.06996279,
        0.08777601,
        -0.04984741,
        -0.09438611,
        -0.08512118,
        0.1115643,
        0.051818553,
        -0.04087223,
        -0.0034000003,
        0.057340942,
        -0.08057381,
        -0.03543507,
        0.10344641,
        -0.006367757,
        -0.046588477,
        -0.014496572,
        -0.00042292883,
        0.08346094,
        -0.009812792,
        0.0022957693,
        -0.06272821,
        0.014676474,
        -0.08511554,
        0.02803771,
        -0.056213416,
        0.1442303,
        0.02647704,
        0.018963736,
        0.01563833,
        -0.012848928,
        0.042879947,
        0.022008887,
        -0.10297819,
        0.014672903,
        0.028854959,
        -0.07638677,
        0.004788842,
        -0.02368985,
        -0.016259484,
        -0.5937118,
        0.007413279,
        0.0005739511,
        -0.02434158,
        -0.12584588,
        0.042361103,
        0.013012061,
        0.0014098343,
        0.07063148,
        -0.00068419793,
        0.093541585,
        -0.04610961,
        0.014931206,
        -0.014423708,
        0.040771093,
        -0.012165938,
        -0.0251508
      ]
    },
    {
      "id": "worldsfair-speaker-314",
      "conference": "worldsfair",
      "name": "Ritvik Pandya",
      "role": "Engineering Leader",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "AI : Learned Execution Graphs for Real-Time Anomaly Detection & Drift Classification in APIs",
          "description": "API ingress controllers process requests through ordered sequences of middleware steps — authentication, authorization, validation, rate limiting, routing, service invocation, caching. We model this pipeline as a directed acyclic graph (DAG) learned from structured telemetry events, then apply graph-based anomaly detection and drift classification in real time at 1,600+ TPS. The system emits one structured event per processing step, constructs per-endpoint execution graphs using sequence mining with statistical confidence thresholds, and learns per-node baselines (latency, dependency, execution frequency). Three graph intelligence capabilities emerge: (1) Graph-based anomaly attribution — compute per-node deviation ratios against learned baselines to identify the exact bottleneck node and its dependency. In production, this pinpointed a 41x deviation at a single graph node that was invisible to service-level monitoring, reducing root cause identification from 2-3 hours to under 30 seconds. (2) Graph structural drift detection — compare observed node sequences against the learned graph topology to detect missing nodes (mandatory processing step silently skipped), reordered nodes (middleware misconfiguration), and unexpected new nodes (unauthorized middleware injection). Traditional monitoring reported \"system healthy\" when a mandatory node was removed — latency dropped, errors at zero — only the learned graph comparison detected the structural change. (3) Per-client graph fingerprinting — learn client-specific execution graph profiles using exponential moving averages. Detect when a client's graph traversal pattern changes, classify the cause (client behavior change vs. configuration drift vs. infrastructure failover) using KL divergence on node-visit distributions, and apply graph-aware adaptive control scoped to specific nodes rather than entire endpoints. The execution graph model also enables a novel approach to retry storm detection: analyzing idempotency key entropy at graph nodes to classify traffic as legitimate growth vs. retry amplification, and returning cached responses at the specific graph node rather than rejecting requests — breaking the retry amplification loop. Production system processing high TPS. Attendees will learn the graph construction methodology, the anomaly attribution algorithm, and concrete patterns for adding learned graph intelligence to any middleware pipeline.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.0048008636,
        0.10435862,
        0.0057755252,
        -0.0010407614,
        0.013487017,
        0.0044407044,
        -0.07160355,
        0.022512097,
        0.10695605,
        -0.22803617,
        -0.005902894,
        0.01823602,
        0.007696235,
        -0.042161547,
        -0.0817972,
        -0.0061079105,
        0.07664454,
        0.0045556636,
        -0.021149926,
        -0.07314721,
        -0.038792305,
        -0.027443627,
        -0.024517423,
        0.02386887,
        -0.036788214,
        0.065902784,
        0.012031441,
        0.04982389,
        -0.08939591,
        0.44297576,
        -0.05852985,
        -0.09377592,
        -0.03860365,
        -0.025289923,
        0.059020035,
        -0.035500117,
        -0.09810664,
        -0.008729482,
        0.045063294,
        0.08757378,
        -0.0071383975,
        0.0667683,
        -0.020425037,
        0.022568775,
        -0.025235485,
        -0.037323873,
        -0.09136047,
        0.032739863,
        -0.005023847,
        -0.015739411,
        -0.085497625,
        -0.0064633503,
        0.0022228605,
        0.0021183454,
        0.022050036,
        0.049059205,
        -0.06287325,
        -0.12698022,
        0.12718432,
        -0.054336272,
        -0.06825467,
        0.039763223,
        -0.007849083,
        0.024137076,
        0.0010224198,
        -0.10299507,
        0.103163145,
        -0.027830157,
        -0.0068476787,
        -0.024886072,
        -0.0066319294,
        0.0030899146,
        -0.08034322,
        0.02680388,
        -0.014982036,
        -0.11259795,
        -0.08479913,
        0.118051976,
        -0.028689211,
        0.01734107,
        -0.006409676,
        0.0055815605,
        -0.11062003,
        0.007839706,
        -0.01282875,
        -0.016466638,
        -0.041258145,
        -0.0049788137,
        -0.023503013,
        0.13625607,
        -0.0015817994,
        -0.036246553,
        -0.021325853,
        -0.05572127,
        -0.034636043,
        0.06828074,
        -0.034986988,
        0.07960871,
        -0.053577673,
        0.06992584,
        0.0182101,
        -0.02007146,
        0.036456585,
        -0.027711201,
        -0.08329683,
        -0.10123889,
        -0.015640754,
        -0.009050319,
        0.021466162,
        0.0018505811,
        -0.06420405,
        -0.5967649,
        0.019895362,
        0.0051748026,
        -0.067154355,
        -0.1610434,
        0.03554283,
        -0.03379857,
        0.082693815,
        0.098920606,
        0.0008180019,
        0.0641615,
        -0.079072915,
        0.0034534195,
        -0.02212584,
        0.06674561,
        -0.007708109,
        0.00762892
      ]
    },
    {
      "id": "worldsfair-speaker-315",
      "conference": "worldsfair",
      "name": "Alex Bauer",
      "role": "Co-Founder, Product",
      "company": "Upside",
      "twitter": "https://x.com/alexdbauer",
      "talks": [
        {
          "title": "The Vibe GTM Iceberg",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03539398,
        0.018553816,
        -0.0122083705,
        -0.025602378,
        -0.026370937,
        -0.049050137,
        -0.10135521,
        0.029962076,
        0.037524175,
        -0.23464137,
        0.0222151,
        -0.046968017,
        0.017982,
        -0.061162062,
        -0.07443912,
        0.047782067,
        0.07765928,
        -0.036684915,
        0.0074771275,
        -0.118182324,
        -0.023101935,
        -0.071050726,
        -0.066191085,
        -0.015408948,
        -0.082892396,
        0.043754514,
        0.018398538,
        0.06336052,
        -0.12948328,
        0.45284125,
        -0.08880614,
        0.022679143,
        -0.048738874,
        0.028632352,
        0.02113554,
        -0.06068602,
        -0.01821134,
        0.043266054,
        -0.031729717,
        0.09659176,
        -0.025374206,
        0.07076863,
        -0.07529492,
        0.028628174,
        -0.0041891076,
        0.054933246,
        -0.061928857,
        -0.027369177,
        -0.05026771,
        -0.034962956,
        -0.061751995,
        0.048166294,
        -0.059356995,
        -0.06976234,
        -0.030778224,
        0.06701897,
        -0.020860352,
        -0.0070126303,
        0.08309837,
        -0.0709834,
        -0.0886476,
        -0.021200556,
        -0.089147985,
        -0.029936368,
        0.02751075,
        -0.013201534,
        0.08714831,
        -0.06194119,
        -0.05122347,
        -0.064336896,
        -0.037510213,
        0.03868377,
        -0.08611924,
        0.092166334,
        -0.091627985,
        -0.049242407,
        -0.031322382,
        0.042069845,
        0.024996016,
        -0.018801088,
        0.060404055,
        0.03812995,
        -0.08419356,
        -0.04606867,
        -0.02245443,
        -0.064399816,
        -0.032771602,
        -0.035191964,
        0.016613415,
        0.0646068,
        -0.003250321,
        0.00016864533,
        -0.014934787,
        0.008311438,
        -0.04174789,
        0.0390784,
        -0.09030026,
        0.08558033,
        -0.025846235,
        0.014513012,
        -0.029077565,
        -0.0777171,
        0.024303596,
        -0.004444676,
        -0.017812029,
        0.004266976,
        0.0067632436,
        0.028904706,
        0.006714283,
        -0.05763242,
        -0.03872359,
        -0.6095313,
        0.09227649,
        0.05490157,
        0.014981858,
        -0.16108008,
        -0.020848941,
        0.030911079,
        0.0046373834,
        0.061621368,
        -0.06593151,
        -0.030362729,
        -0.038280085,
        -0.014553664,
        0.047924507,
        0.07016958,
        0.057201244,
        0.028006349
      ]
    },
    {
      "id": "worldsfair-speaker-316",
      "conference": "worldsfair",
      "name": "Clay Cockrell",
      "role": "Licensed Clinical Social Worker (LCSW); Founder",
      "company": "CoupleWork",
      "twitter": "",
      "talks": [
        {
          "title": "AI is the becoming the World's largest Relationship Therapist. We Can't Afford to Get it Wrong.",
          "description": "Millions of people are now turning to AI for relationship advice and emotional support, often before they'd ever consider a human therapist. Most of the AI Therapy that is available is without clinical oversight, ethical frameworks, or any serious reckoning with what it means to intervene in the most intimate and vulnerable space in a person's life. People are getting hurt. As a couples therapist with 30 years experience, I teamed up with the former CTO at S&P and we created CoupleWork, an AI relationship therapist I essentially trained on three decades of clinical knowledge and every evidence-based modality that exists. Our voice interactive AI, Maxine, is proving this can be done responsibly and very effectively. And what we're learning about the nature of love, connection, and human vulnerability at scale is something this industry needs to hear. I also want to talk about what comes next: the regulatory frameworks that don't yet exist, the liability questions nobody is answering, and why the therapists who should be leading this conversation are almost entirely absent from it.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03220316,
        0.07415131,
        0.011456423,
        -0.015660226,
        0.0071358583,
        -0.015093122,
        -0.09630024,
        -0.06594005,
        0.01766307,
        -0.13541555,
        -0.031494934,
        -0.07668639,
        -0.036497533,
        -0.063613065,
        -0.034206267,
        0.06451071,
        0.073960505,
        -0.03210197,
        -0.06467027,
        -0.06720684,
        -0.1253154,
        -0.004892786,
        -0.030688416,
        0.05910654,
        -0.04247654,
        0.14047338,
        0.014461405,
        0.03573579,
        -0.11305701,
        0.45069894,
        -0.09826514,
        0.04114024,
        0.024166433,
        0.04111955,
        0.030055495,
        -0.031061493,
        -0.0012052207,
        -0.0055571897,
        0.051662333,
        -0.037443466,
        -0.003675466,
        0.01620172,
        -0.02207447,
        0.017480496,
        0.040146593,
        -0.06131747,
        -0.048990097,
        -0.021980012,
        0.043492384,
        -0.0325973,
        -0.02389217,
        -0.0654389,
        0.07423528,
        0.00028756016,
        0.01967908,
        0.035325665,
        0.029361507,
        -0.07644357,
        0.07265697,
        -0.059740312,
        -0.16772799,
        0.020210717,
        -0.021209281,
        0.06160378,
        -0.0010081114,
        -0.079416186,
        -0.01686944,
        -0.063258365,
        -0.06468623,
        -0.019337637,
        0.01247782,
        -0.025566543,
        -0.058770064,
        -0.08783407,
        -0.10101243,
        -0.05548454,
        -0.025829203,
        0.06992762,
        -0.00018605028,
        -0.038169887,
        0.015955215,
        0.053942148,
        -0.13604027,
        -0.037952267,
        -0.009335694,
        -0.07186371,
        -0.044048253,
        0.048437953,
        -0.058902483,
        0.086172156,
        0.040285494,
        -0.030008044,
        -0.06763243,
        -0.008909364,
        -0.07387509,
        0.010634385,
        -0.049339376,
        0.04229303,
        -0.0357973,
        0.013144999,
        0.04470408,
        -0.058615994,
        0.030089859,
        -0.017790567,
        -0.061303545,
        -0.08320095,
        -0.007871309,
        -0.00769042,
        0.15927307,
        0.083583854,
        -0.03661359,
        -0.60025233,
        0.00815872,
        0.050398067,
        -0.036452275,
        -0.12850834,
        0.049988654,
        0.02978693,
        0.0030593984,
        0.07135104,
        -0.11201523,
        0.030201716,
        0.0454846,
        0.02172357,
        0.004311712,
        0.089862145,
        0.03409403,
        0.037441924
      ]
    },
    {
      "id": "worldsfair-speaker-317",
      "conference": "worldsfair",
      "name": "Tony Fabrikant",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "AI is the becoming the World's largest Relationship Therapist. We Can't Afford to Get it Wrong.",
          "description": "Millions of people are now turning to AI for relationship advice and emotional support, often before they'd ever consider a human therapist. Most of the AI Therapy that is available is without clinical oversight, ethical frameworks, or any serious reckoning with what it means to intervene in the most intimate and vulnerable space in a person's life. People are getting hurt. As a couples therapist with 30 years experience, I teamed up with the former CTO at S&P and we created CoupleWork, an AI relationship therapist I essentially trained on three decades of clinical knowledge and every evidence-based modality that exists. Our voice interactive AI, Maxine, is proving this can be done responsibly and very effectively. And what we're learning about the nature of love, connection, and human vulnerability at scale is something this industry needs to hear. I also want to talk about what comes next: the regulatory frameworks that don't yet exist, the liability questions nobody is answering, and why the therapists who should be leading this conversation are almost entirely absent from it.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.02656898,
        0.03288329,
        -0.04104555,
        -0.0132993115,
        -0.00044014087,
        -0.020373967,
        -0.117056474,
        -0.02968241,
        0.041410547,
        -0.21735537,
        -0.03657318,
        -0.019732188,
        -0.05485468,
        -0.032751787,
        0.0023370888,
        0.02575796,
        0.074593455,
        -0.1031081,
        -0.014677308,
        -0.082807675,
        -0.07138622,
        -0.078962326,
        -0.007669484,
        0.06170908,
        -0.03218912,
        0.1825815,
        0.004940623,
        0.051428255,
        -0.08259753,
        0.42520615,
        -0.032180198,
        0.07672165,
        0.03570299,
        -0.012440654,
        0.0077627846,
        0.01412019,
        -0.032108583,
        -0.02334019,
        0.030362435,
        0.045952003,
        0.025270615,
        0.020699108,
        0.031708922,
        -0.027180944,
        -0.030229805,
        -0.07533529,
        -0.10070833,
        -0.06756261,
        0.0141354585,
        -0.047451723,
        -0.02198655,
        -0.0094255805,
        -0.0052688997,
        0.0017050061,
        0.03903253,
        -0.0029392117,
        0.046087988,
        -0.040873706,
        0.06488234,
        -0.024776055,
        -0.11000392,
        -0.015552194,
        -0.05242185,
        0.0741888,
        0.009942619,
        -0.07655201,
        0.036838472,
        -0.1084477,
        -0.008996923,
        -0.062277045,
        -0.031115415,
        -0.049565274,
        -0.059081063,
        -0.018552458,
        -0.11228121,
        -0.08662442,
        -0.025304925,
        0.08852022,
        0.03348567,
        -0.02241922,
        0.011887947,
        0.069768205,
        -0.10528602,
        -0.076727316,
        0.012482938,
        -0.083506696,
        -0.006907696,
        0.001513116,
        -0.06723899,
        0.024586849,
        0.05264873,
        -0.025028558,
        -0.049480047,
        0.018220669,
        -0.029339882,
        0.048760936,
        -0.04598577,
        0.059308767,
        0.014650182,
        0.00015700994,
        0.038763583,
        -0.022931611,
        0.07444098,
        0.066558085,
        -0.017167918,
        -0.045126017,
        -0.042964168,
        -0.00849532,
        0.16139928,
        0.0530379,
        -0.028834313,
        -0.6231533,
        0.00594945,
        -0.009699637,
        -0.111250095,
        -0.11774914,
        0.025866156,
        0.016025294,
        -0.011961783,
        0.015198724,
        -0.05656264,
        0.05052447,
        0.04477494,
        -0.025260374,
        0.023587078,
        0.08570288,
        0.057867996,
        -0.0048763747
      ]
    },
    {
      "id": "worldsfair-speaker-318",
      "conference": "worldsfair",
      "name": "Philip Kiely",
      "role": "Developer Relations",
      "company": "Baseten",
      "twitter": "",
      "talks": [
        {
          "title": "What's New in Inference Engineering",
          "description": "More than 30,000 engineers have learned the fundamentals of inference since Inference Engineering was published. But the field keeps accelerating, so it's time for the first public addendum to the book. The past four months have seen a renewed focus on training-dependent inference optimization across the \"big three\" performance techniques of speculation, caching, and quantization. This talk provides structured guidance for training DFlash and EAGLE 3 draft models to accelerate LLM decode, introduces the concept of KV compaction, and explains the hype behind TurboQuant.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.065424345,
        0.047075383,
        0.03981243,
        0.07611607,
        -0.003982595,
        -0.005243462,
        -0.08377207,
        -0.0022999677,
        0.03429383,
        -0.19698152,
        -0.02529028,
        -0.038258836,
        -0.035118107,
        -0.06404221,
        -0.04569634,
        -0.013824846,
        0.1251035,
        -0.094227165,
        -0.022108149,
        -0.08079095,
        -0.03442838,
        -0.13480087,
        0.010949967,
        0.0951465,
        -0.019832937,
        0.099408016,
        0.025959322,
        0.093362235,
        -0.13429846,
        0.46969652,
        -0.13322963,
        -0.0019218656,
        -0.002250412,
        -0.024271242,
        0.05619052,
        -0.043031577,
        -0.034267906,
        0.03492329,
        -0.03129714,
        0.020811766,
        0.017408673,
        -0.03394673,
        -0.07396152,
        0.02352514,
        0.00013823604,
        -0.009468131,
        -0.08292648,
        -0.0007429407,
        -0.03722362,
        -0.003320233,
        -0.039573967,
        0.010034178,
        0.04530744,
        0.060265902,
        0.054191735,
        0.04534222,
        0.00552585,
        -0.07198367,
        0.11666722,
        -0.013144387,
        -0.08774973,
        0.011260533,
        -0.008077497,
        -0.010812629,
        -0.025526885,
        -0.08599733,
        0.02691884,
        -0.03143493,
        -0.03255771,
        0.012853941,
        0.050485145,
        -0.02351301,
        -0.05304883,
        0.0026535015,
        -0.04056963,
        -0.0895649,
        -0.033821903,
        0.040897112,
        0.03674465,
        0.015535829,
        0.0034098828,
        0.04804074,
        -0.0024025165,
        -0.07215161,
        0.11246204,
        -0.048035774,
        -0.017011922,
        0.029919026,
        -0.010009331,
        0.049786393,
        0.037282575,
        -0.032998838,
        -0.034426358,
        -0.039197642,
        -0.04345214,
        0.0379998,
        0.04028777,
        0.03474347,
        -0.018286422,
        0.03038887,
        0.021158986,
        -0.004110006,
        0.004443857,
        -0.06217123,
        -0.06945209,
        -0.06546747,
        0.031897604,
        0.008618651,
        0.06118481,
        -0.0068314853,
        -0.05658568,
        -0.61974156,
        0.041883327,
        -0.04594575,
        -0.022121383,
        -0.12729867,
        0.053904936,
        -0.09249806,
        0.00867815,
        0.097647265,
        -0.027889624,
        0.069357015,
        -0.02862305,
        -0.002150928,
        0.026543986,
        0.04263925,
        0.04701997,
        0.019019747
      ]
    },
    {
      "id": "worldsfair-speaker-319",
      "conference": "worldsfair",
      "name": "Josh Leavitt",
      "role": "Senior Director of Product Management",
      "company": "Coinbase",
      "twitter": "",
      "talks": [
        {
          "title": "From Zero to AI-Native: Scaling AI Across the Org",
          "description": "Most companies talk about being AI-native, but few show what it takes at scale. Josh Leavitt, Sr. Director of AI & Data at Coinbase, shares the hard-won playbook for transforming a high-stakes, regulated engineering organization into one where AI is a first-class citizen across every team. Josh can cover my approach towards building a centralized AI platform that serves thousands of engineers without becoming a bottleneck, tooling decisions that actually moved the needle, and what AI-native really means when shipping in a zero-tolerance-for-failure environment. Expect concrete frameworks, real examples, and honest lessons from what didn’t work.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.032034412,
        0.06728277,
        -0.04711156,
        0.023074802,
        -0.033681877,
        -0.05278845,
        -0.15590617,
        0.021427637,
        0.053609762,
        -0.24363023,
        -0.054637924,
        -0.04809728,
        -0.028391473,
        -0.051316045,
        -0.01708981,
        0.03801727,
        0.09201002,
        -0.042229358,
        -0.06449909,
        -0.026168616,
        -0.04290038,
        0.01808363,
        -0.019938728,
        0.045974147,
        -0.021260042,
        0.10172153,
        0.08746026,
        0.0663121,
        -0.08577513,
        0.45152572,
        -0.08117869,
        -0.00090805005,
        -0.032053847,
        -0.002754505,
        0.03600239,
        -0.022568991,
        -0.07773031,
        -0.033555895,
        -0.014107652,
        0.056209173,
        -0.004916468,
        0.048252948,
        -0.037440006,
        -0.014277379,
        -0.0016649397,
        -0.048751142,
        -0.056900278,
        -0.03824095,
        0.018225618,
        -0.03203471,
        0.0038380602,
        -0.008366514,
        0.031957913,
        0.0035559072,
        0.010949537,
        0.07910819,
        0.02217872,
        -0.07568278,
        0.0796413,
        -0.05628874,
        -0.10533996,
        0.036920156,
        0.0029812907,
        -0.07780636,
        0.05626071,
        -0.11835364,
        0.043027088,
        -0.030307986,
        -0.043502294,
        0.0369823,
        0.024402577,
        0.0630614,
        -0.06315444,
        -0.03372914,
        -0.08945352,
        -0.11747895,
        0.012546871,
        0.103947066,
        -0.03562479,
        0.054302663,
        0.026301384,
        -0.011214643,
        -0.082397774,
        -0.0021042796,
        0.069357805,
        -0.008032724,
        -0.075193346,
        0.04659225,
        -0.027782295,
        0.098403886,
        0.023214556,
        -0.0063166087,
        -0.03758138,
        -0.011514039,
        -0.11630801,
        0.018725399,
        -0.032840423,
        0.026995093,
        -0.034077175,
        0.062625304,
        0.023286643,
        -0.05848556,
        0.058990814,
        0.027475176,
        -0.0030244633,
        -0.04921792,
        0.004326849,
        0.023506274,
        0.034576103,
        0.016980598,
        -0.05460698,
        -0.6118275,
        0.0141645465,
        0.04724348,
        -0.044409,
        -0.14361785,
        0.05749182,
        -0.0035925433,
        0.009580947,
        0.114686765,
        -0.041217715,
        -0.0038656606,
        -0.03658031,
        0.0040499195,
        0.026782297,
        -0.0011968599,
        0.05811617,
        -0.006808939
      ]
    },
    {
      "id": "worldsfair-speaker-320",
      "conference": "worldsfair",
      "name": "Patrick Debois",
      "role": "Product Developer Relations Lead",
      "company": "Tessl",
      "twitter": "https://x.com/patrickdebois",
      "talks": [
        {
          "title": "Coding Agents Don't Scale Themselves. Neither Do Your Teams.The Rise of Agent Enablement.",
          "description": "Every company wants to know how others are actually scaling AI coding. But it's hard to get past the generic transformation stories. What are the new practices showing up in real engineering orgs? What does maturity actually look like, and what separates teams that are moving from teams that are stuck? What are the patterns for enabling humans and agents, together? Patrick Debois has been collecting the practices and patterns, talking to the early Agent Enablement teams already on the job, team leads, and VPs of Engineering. What's showing up is a new function: a team that enables other teams to get real leverage out of their agents. This talk takes the [Context Development Lifecycle](https://tessl.io/blog/context-development-lifecycle-better-context-for-ai-coding-agents/) off the individual laptop and onto the org chart, grouped across three pillars: - **Enablement.** From individual experimentation to team and org-level fluency with agents. - **Platform.** Agent tooling that runs like a real delivery pipeline: fast, observable, cost-aware. - **Governance.** Ad-hoc guardrails growing into real evaluation, telemetry, and accountable agent work. For Agent Enablement leaders scaling it out across the org. For team leads looking to help their teams get better at this. For VPs ready to unblock the friction and unlock what agents can actually do. *Coding agents don't scale themselves. This is the talk about who does*",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.026068307,
        0.048011556,
        0.025626235,
        0.0037643511,
        -0.008144801,
        -0.06005496,
        -0.09303437,
        -0.010718685,
        0.027504954,
        -0.24166514,
        -0.07489501,
        -0.050581522,
        -0.053313192,
        -0.06121917,
        -0.014918656,
        -0.077709965,
        0.07602623,
        -0.07301505,
        -0.03332126,
        -0.025154196,
        0.0049338825,
        -0.01519731,
        -0.03886377,
        0.08725023,
        -0.019681789,
        0.16076216,
        -0.008594125,
        0.069770716,
        -0.077355854,
        0.45655492,
        -0.1219799,
        0.08281582,
        0.006718365,
        -0.024128363,
        0.031060126,
        -0.011401611,
        -0.03578032,
        -0.011404877,
        0.0110706035,
        0.048420835,
        -0.010678888,
        0.06722003,
        -0.05140742,
        0.019791225,
        -0.0705768,
        -0.009822553,
        -0.1553119,
        0.082675725,
        0.020795459,
        -0.032273255,
        0.0074142977,
        -0.0012551136,
        0.09501998,
        -0.015213364,
        0.05951779,
        0.015074191,
        0.012064327,
        -0.07452797,
        0.09143154,
        0.02566279,
        -0.07481773,
        -0.010853883,
        -0.00018990974,
        -0.010156033,
        -0.096013516,
        -0.10894856,
        0.055217702,
        -0.037722204,
        -0.014732258,
        -0.021067854,
        0.03906609,
        0.07044113,
        -0.07237512,
        0.040238053,
        -0.07495582,
        -0.14073567,
        -0.049088225,
        0.1032227,
        -0.058678113,
        0.0038940408,
        -0.082137324,
        0.05992829,
        -0.04932214,
        -0.04946844,
        0.030720744,
        0.029123422,
        -0.092902206,
        -0.016895702,
        -0.027114177,
        0.05185846,
        0.027743958,
        0.009238382,
        -0.016150432,
        0.026966777,
        -0.10697739,
        0.016603362,
        -0.021252066,
        -0.011123988,
        0.0018517919,
        -0.016971897,
        -0.05780709,
        -0.076813884,
        0.012452094,
        -0.041943308,
        0.045805853,
        0.004703503,
        -0.03388674,
        0.01914322,
        0.04874641,
        0.059075568,
        0.0030571918,
        -0.5699833,
        -0.018640123,
        0.02080954,
        -0.03929303,
        -0.10154336,
        0.045421716,
        -0.045985337,
        0.057961427,
        0.123057276,
        -0.025779836,
        0.079224326,
        -0.08498777,
        -0.020476611,
        0.017240359,
        0.09119736,
        0.017826416,
        -0.0090160975
      ]
    },
    {
      "id": "worldsfair-speaker-321",
      "conference": "worldsfair",
      "name": "Eric Jia",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "YOLO Mode, Safely: microVM Sandboxes for Any Agent",
          "description": "This talk shows the alternative: every agent session in its own microVM, with its own kernel and a hard boundary to the host. You decide what lives inside that boundary: filesystem, network, the tools it's allowed to call. The sandbox runs Claude Code, Cursor, Codex, or whatever else you're driving. You'll see an agent live in full YOLO mode inside a sandbox, a real attempt to escape, and the boundary that holds up.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.069867715,
        0.068433665,
        -0.008813486,
        0.0025045336,
        0.048472863,
        0.0070638056,
        -0.09714123,
        -0.041434247,
        0.026099738,
        -0.22974996,
        -0.020717215,
        0.027259413,
        0.0635321,
        -0.03365659,
        -0.035290837,
        0.079829834,
        0.05649801,
        -0.104666755,
        -0.05032462,
        -0.0630036,
        -0.0621038,
        -0.11885033,
        0.0036848823,
        -0.058593504,
        -0.04706009,
        0.12387989,
        -0.008386708,
        0.08876905,
        -0.09667775,
        0.41975737,
        -0.03692004,
        -0.009654878,
        0.031526934,
        -0.024532095,
        0.026021782,
        -0.08542294,
        -0.012799952,
        0.030172648,
        0.06289124,
        0.01743378,
        0.06753154,
        0.01573126,
        -0.0459541,
        0.020423776,
        -0.05589928,
        -0.008097267,
        -0.08422096,
        0.044114206,
        -0.028226804,
        0.015809588,
        -0.04868562,
        0.052734923,
        0.09975912,
        -0.086987734,
        0.015434153,
        -0.011950133,
        -0.02862739,
        -0.029581498,
        0.089810096,
        0.002281909,
        -0.1008974,
        -0.04331624,
        -0.017399395,
        0.021069784,
        -0.05626732,
        -0.00020031066,
        0.07106065,
        -0.07751728,
        0.018443458,
        -0.060089353,
        0.04206089,
        0.046787433,
        -0.010237119,
        0.022459555,
        -0.10203728,
        -0.11535685,
        -0.006791637,
        0.02484682,
        -0.04833171,
        -0.0194783,
        -0.03482017,
        0.097945586,
        -0.08130756,
        -0.055617772,
        -0.023308037,
        0.000026253048,
        -0.06631822,
        0.06615997,
        -0.062420893,
        0.078165,
        0.017295161,
        0.05233585,
        -0.084059864,
        0.04219531,
        -0.053637784,
        0.026302313,
        -0.04311703,
        -0.0042698616,
        -0.01571506,
        0.024923809,
        0.014157979,
        0.029197972,
        0.08492522,
        0.0062990407,
        -0.020251552,
        -0.037846763,
        0.017224466,
        0.0048816213,
        0.0666031,
        0.015248693,
        -0.023706218,
        -0.6105953,
        -0.067242235,
        -0.00097419275,
        -0.10223112,
        -0.15276484,
        0.03531908,
        0.051208995,
        0.072276644,
        0.06970927,
        -0.09527384,
        0.1011457,
        -0.0002969619,
        -0.016893698,
        -0.05491886,
        0.032034006,
        -0.011202426,
        0.034982808
      ]
    },
    {
      "id": "worldsfair-speaker-322",
      "conference": "worldsfair",
      "name": "Joshua Mo",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Your Model is Private. Your System Isn't.",
          "description": "Privacy in AI isn't just about choosing the right model. Data leaks rarely happen inside the LLM itself - they happen in the systems surrounding it. Observability pipelines, analytics platforms, prompts, agents, and infrastructure often become accidental channels for exposing user data. In this session, Joshua Mo, Lead DevRel Engineer at Venice AI, explores why private models alone are not enough and shares practical privacy-preserving patterns that AI engineers can adopt today. From revocable handles and hashed identifiers to agent boundaries and confidential computing, attendees will leave with concrete ideas for building AI systems that protect user data by design.",
          "day": "Day 4 — Session Day 3",
          "time": "1:30pm-1:50pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04975724,
        0.051706057,
        0.012193288,
        0.022199845,
        0.05603083,
        -0.056302,
        -0.038316585,
        0.028071385,
        0.021727894,
        -0.25525442,
        -0.06015098,
        -0.0328429,
        -0.030989144,
        -0.037992798,
        -0.04392293,
        0.025248397,
        0.04920875,
        -0.035027668,
        -0.09308919,
        -0.06938153,
        -0.032607723,
        -0.1152668,
        -0.0015682459,
        0.03622084,
        -0.06851054,
        0.058285322,
        0.01625923,
        0.06107898,
        -0.11214378,
        0.4098704,
        -0.08600621,
        0.03956174,
        -0.016909078,
        -0.04379417,
        0.06003639,
        -0.01716508,
        -0.04634777,
        0.006449399,
        0.0317945,
        0.08672711,
        -0.0036194613,
        0.016987657,
        -0.039862603,
        0.090577245,
        -0.021514943,
        -0.06032649,
        -0.106971525,
        0.04171267,
        0.032491628,
        -0.034195088,
        -0.06428184,
        -0.034428127,
        0.07849925,
        -0.017724948,
        -0.0027833139,
        0.050521154,
        0.021842435,
        -0.0045899726,
        0.12197713,
        -0.03507956,
        -0.16034189,
        -0.061709452,
        -0.018284384,
        -0.019046564,
        -0.038530782,
        -0.060659945,
        0.050992694,
        -0.07230298,
        -0.013981078,
        -0.030924056,
        0.005487829,
        0.0073786112,
        -0.046718623,
        0.03825898,
        -0.106831826,
        -0.13265705,
        0.0027021542,
        0.044083446,
        -0.03678749,
        0.012743521,
        0.0071223537,
        0.06910429,
        -0.028562546,
        -0.108424634,
        0.040047586,
        -0.013157091,
        -0.0902062,
        0.03505231,
        -0.023823503,
        0.11757197,
        -0.016507367,
        0.048208013,
        -0.053760305,
        0.041694485,
        -0.043981563,
        0.012278829,
        -0.044867538,
        0.020772206,
        -0.025338681,
        -0.020961067,
        0.045447383,
        -0.014750621,
        0.0036343911,
        0.058365054,
        -0.006032923,
        -0.011966898,
        -0.008491659,
        0.00008494634,
        0.110917896,
        0.040303323,
        -0.078534186,
        -0.62209344,
        -0.0138464235,
        -0.0018163772,
        -0.04730237,
        -0.121744886,
        0.020605046,
        0.0103557175,
        0.06617537,
        0.1399056,
        -0.057449132,
        0.017610824,
        0.027492933,
        0.007651006,
        -0.023288114,
        0.048491053,
        0.008964317,
        -0.005132402
      ]
    },
    {
      "id": "worldsfair-speaker-323",
      "conference": "worldsfair",
      "name": "Giselle van Dongen",
      "role": "",
      "company": "Restate",
      "twitter": "https://x.com/vdgiselle",
      "talks": [
        {
          "title": "🎵 Every step you take, every call you make - the reliable agent stack",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04440808,
        -0.0044659125,
        0.01801232,
        0.0277022,
        0.05571693,
        0.0025252805,
        -0.04291689,
        0.0010559411,
        -0.011580177,
        -0.15280968,
        -0.0011861678,
        -0.06426811,
        -0.000029785724,
        -0.0020757306,
        -0.032620434,
        -0.0040997835,
        0.10345066,
        -0.060144756,
        -0.110562906,
        -0.16398056,
        -0.014644625,
        -0.05564587,
        -0.005538524,
        0.014667487,
        -0.034207273,
        0.10488308,
        -0.015263571,
        0.013256892,
        -0.07026384,
        0.5110076,
        -0.11314388,
        0.043873575,
        0.033801578,
        -0.033732846,
        0.09927682,
        -0.09827037,
        0.007664312,
        -0.03175594,
        0.038218953,
        0.07996683,
        0.008143489,
        0.02362365,
        -0.03988078,
        0.03986519,
        -0.09732415,
        0.044017516,
        -0.058329556,
        -0.0035423555,
        0.025276976,
        0.005967487,
        -0.10348569,
        0.0019379501,
        0.051946875,
        -0.04604808,
        0.06777214,
        -0.013212826,
        0.012704763,
        -0.0769532,
        0.048969917,
        -0.02605944,
        -0.056091037,
        0.015659533,
        -0.042297307,
        0.0043003615,
        0.031513512,
        -0.016690714,
        0.0016630644,
        -0.047014084,
        0.032397255,
        -0.06779298,
        -0.025290439,
        -0.06592526,
        -0.035579357,
        0.026988203,
        -0.048736487,
        -0.07808142,
        -0.005174352,
        0.04619065,
        -0.047886647,
        0.000055994737,
        -0.023221353,
        0.06677278,
        -0.12264045,
        -0.021007981,
        0.06343829,
        -0.028023813,
        -0.00656915,
        0.020919908,
        -0.10584247,
        -0.015535636,
        0.03797261,
        -0.017635165,
        0.024587084,
        0.04083661,
        -0.08762765,
        0.036970694,
        -0.050081175,
        0.046200745,
        0.026406787,
        -0.005926306,
        -0.060753368,
        -0.054746278,
        0.018262532,
        -0.02819564,
        -0.024858952,
        -0.031495754,
        -0.043375522,
        -0.04989041,
        0.056793645,
        -0.010806551,
        -0.051476106,
        -0.6001103,
        -0.041883416,
        -0.03525884,
        -0.015399589,
        -0.17755297,
        -0.055279497,
        0.029468928,
        0.008740315,
        0.097850956,
        -0.023781585,
        0.06468222,
        -0.050160255,
        0.025981544,
        0.027810093,
        0.057963725,
        0.041029543,
        -0.03888042
      ]
    },
    {
      "id": "worldsfair-speaker-324",
      "conference": "worldsfair",
      "name": "Lina Colucci",
      "role": "Co-founder and CEO",
      "company": "Lemon Slice",
      "twitter": "https://x.com/lina_colucci",
      "talks": [
        {
          "title": "Voice agents with Realtime Video",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.026288435,
        0.0073339646,
        0.0035144235,
        0.023279134,
        0.044381548,
        -0.026959378,
        -0.054884654,
        0.04141443,
        0.055352155,
        -0.17652163,
        0.044788554,
        -0.007135264,
        -0.021720687,
        -0.029302865,
        -0.044412002,
        0.05732266,
        0.105076395,
        -0.07726764,
        -0.07325171,
        -0.11206843,
        -0.07904977,
        -0.07053093,
        0.004367115,
        0.00092639314,
        -0.037360225,
        0.13536274,
        0.009156667,
        -0.055028617,
        -0.132379,
        0.49208757,
        -0.096674494,
        -0.0018478503,
        0.008962852,
        -0.04148926,
        0.059075575,
        -0.01024621,
        0.002610245,
        0.0075850855,
        0.025671486,
        0.08557637,
        -0.03512288,
        0.048929136,
        -0.0066764476,
        0.01045794,
        -0.006149383,
        -0.034614358,
        -0.04690123,
        0.025448214,
        -0.0010237334,
        0.024080012,
        -0.023698837,
        -0.008162237,
        -0.0066784625,
        -0.054542866,
        0.07091248,
        -0.058851887,
        -0.023632959,
        -0.040929683,
        0.087975524,
        -0.0337038,
        -0.090002306,
        -0.058346473,
        -0.03537438,
        0.0104657365,
        0.010399807,
        -0.041156277,
        0.03634577,
        -0.10096064,
        -0.019279093,
        -0.10666618,
        0.0060800416,
        0.032759298,
        -0.045166545,
        0.004051339,
        -0.003973817,
        -0.051250968,
        -0.016300162,
        0.09333776,
        0.016192913,
        -0.0060056746,
        -0.007401618,
        0.06356981,
        -0.05539877,
        -0.03258171,
        0.016787533,
        -0.072426565,
        -0.10322707,
        0.06218701,
        -0.057915337,
        0.06964759,
        0.070554934,
        0.008471612,
        -0.038972266,
        -0.027235309,
        -0.05392333,
        0.032864317,
        -0.07609548,
        0.03538963,
        0.044868264,
        0.011071436,
        -0.035086215,
        -0.07091329,
        0.008226152,
        0.034835603,
        -0.025118696,
        -0.020040814,
        0.013395636,
        -0.006954646,
        0.10988011,
        -0.085015014,
        -0.103551656,
        -0.6097068,
        0.03582379,
        -0.025256153,
        -0.0022011527,
        -0.17374925,
        -0.02752746,
        0.0026376373,
        0.048158333,
        0.035015736,
        -0.04323315,
        0.052232996,
        -0.01688765,
        0.025014078,
        0.042843454,
        0.018407712,
        0.014350033,
        0.019452982
      ]
    },
    {
      "id": "worldsfair-speaker-325",
      "conference": "worldsfair",
      "name": "Anna Spysz",
      "role": "",
      "company": "Stripe",
      "twitter": "",
      "talks": [
        {
          "title": "Teaching agents to pay",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.057392478,
        -0.004129142,
        0.058287293,
        -0.0055036964,
        0.028691351,
        -0.05248337,
        -0.05578247,
        -0.02463509,
        0.026375214,
        -0.24949962,
        -0.035550125,
        -0.038132217,
        0.021652889,
        -0.008084004,
        -0.056750234,
        0.018425362,
        0.06471096,
        -0.08719712,
        -0.047587004,
        -0.036606573,
        0.0017736545,
        -0.115961075,
        -0.009420638,
        0.034761496,
        -0.022640817,
        0.12819882,
        -0.07934613,
        0.03226353,
        -0.08242518,
        0.5000032,
        -0.05686078,
        -0.017729767,
        0.0072211307,
        0.0101473695,
        0.08613381,
        -0.06652679,
        -0.057323884,
        -0.08406293,
        0.051324047,
        0.0962117,
        0.007343621,
        -0.022919882,
        -0.06697703,
        0.006687365,
        -0.019381095,
        -0.019816507,
        -0.067783944,
        -0.033116024,
        -0.02658537,
        0.026071817,
        -0.015006248,
        0.029596243,
        -0.008983678,
        -0.033965398,
        0.07910579,
        0.017211966,
        -0.031163404,
        -0.06811338,
        0.15014179,
        -0.07756788,
        -0.06519699,
        -0.015145216,
        -0.042625196,
        0.064352855,
        0.051744044,
        -0.029074261,
        0.019655202,
        -0.0952716,
        0.048374385,
        -0.09663687,
        -0.013835477,
        0.034486093,
        -0.0014420532,
        0.080104895,
        -0.04306371,
        -0.047430854,
        0.0012710693,
        0.053618353,
        -0.00961482,
        0.083778724,
        0.04901216,
        0.030656997,
        -0.041790973,
        -0.038312543,
        0.041128036,
        -0.051641256,
        0.023587111,
        0.0025226753,
        -0.026132492,
        -0.0049165175,
        0.012157891,
        -0.04152804,
        0.0042889025,
        -0.02841814,
        -0.011240931,
        -0.006371685,
        -0.14673017,
        0.0881138,
        0.043815788,
        0.008942663,
        -0.014584709,
        -0.055463627,
        0.033971272,
        -0.020580666,
        -0.0125200115,
        0.0032390007,
        -0.0043704477,
        -0.024893004,
        0.08118401,
        -0.030414358,
        -0.033747803,
        -0.5815356,
        -0.057778012,
        -0.017134512,
        -0.023689156,
        -0.15577276,
        -0.03863934,
        -0.037548374,
        0.0017276654,
        0.10206394,
        -0.04042491,
        0.060333084,
        -0.062374674,
        -0.03251038,
        -0.00057780143,
        0.021688728,
        0.04914413,
        0.020966262
      ]
    },
    {
      "id": "worldsfair-speaker-326",
      "conference": "worldsfair",
      "name": "Lucas Palma",
      "role": "Information Security Manager",
      "company": "Nubank",
      "twitter": "",
      "talks": [
        {
          "title": "We Vetted 2,000 AI Skills Before They Reached Developers",
          "description": "AI skills and plugins are becoming part of the software supply chain. They steer agent behavior, describe tools, run commands, access files, and shape how developers build with AI. Treating them as harmless configuration is a mistake. This talk shares what we learned from building an automated security review system for more than 2,000 internal AI skills before they reached a company wide plugin marketplace. I will walk through the risks we found, the checks that worked, the checks that created noise, and how we turned skill review into something developers could run locally and in CI. We will cover practical patterns for reviewing unsafe instructions, destructive commands, sensitive data exposure, risky tool use, credential handling, external calls, and agent behavior drift. The goal is to help AI engineers think about skills, plugins, and agent instructions as production dependencies that deserve review before they reach real users.",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.086856276,
        0.036942955,
        -0.0152650615,
        0.003147043,
        0.024689866,
        -0.05332779,
        -0.02955145,
        0.016431661,
        0.050024178,
        -0.18843545,
        -0.010199309,
        -0.031203361,
        0.03551056,
        -0.020824555,
        -0.05561151,
        -0.022223122,
        0.11804211,
        -0.05257726,
        -0.056929663,
        -0.06784211,
        -0.078986086,
        -0.090382576,
        0.0018923471,
        0.05858295,
        -0.06656785,
        0.10640442,
        0.1061666,
        0.024227802,
        -0.08247038,
        0.4873367,
        -0.06674395,
        0.03888848,
        0.0296995,
        -0.058363028,
        0.058666155,
        -0.079291835,
        -0.03046074,
        -0.0066729994,
        -0.04124618,
        -0.002831121,
        -0.011413905,
        0.055460837,
        -0.06755843,
        0.036112055,
        -0.0152895795,
        -0.067013554,
        -0.041562926,
        0.035204165,
        0.0709223,
        -0.042737745,
        -0.04743299,
        0.015038782,
        0.07682133,
        0.06942081,
        0.050755467,
        -0.022713196,
        0.039167725,
        -0.039567955,
        0.09835553,
        -0.0014063952,
        -0.113466196,
        0.036943153,
        0.006644118,
        0.026580507,
        0.018254913,
        -0.104158156,
        0.072199576,
        -0.03655559,
        0.018196518,
        -0.05310969,
        0.01859319,
        0.022359343,
        -0.02283762,
        0.019545957,
        -0.063827634,
        -0.123164825,
        -0.025956959,
        0.015624321,
        -0.03024571,
        0.034421396,
        0.011769756,
        0.006339401,
        -0.07228673,
        -0.023633039,
        0.015700158,
        0.0037135407,
        -0.05699927,
        0.082154,
        -0.01366175,
        0.12239607,
        -0.0020389622,
        0.021283694,
        -0.026526034,
        0.05794138,
        -0.03572953,
        0.018134667,
        -0.012727388,
        0.043804605,
        0.018470429,
        0.03257333,
        0.022701766,
        -0.042080082,
        0.0036224993,
        -0.0338649,
        0.060535885,
        -0.07008552,
        -0.031767674,
        0.023108901,
        0.10215157,
        -0.03134854,
        -0.001934695,
        -0.5997668,
        0.012418652,
        -0.05779948,
        -0.033096753,
        -0.13752088,
        0.04598193,
        -0.0063878926,
        0.038642135,
        0.14175412,
        -0.046531532,
        0.06083138,
        -0.09484897,
        -0.021490654,
        -0.019116372,
        0.04744942,
        0.019693071,
        -0.032788936
      ]
    },
    {
      "id": "worldsfair-speaker-327",
      "conference": "worldsfair",
      "name": "Arjun Singh",
      "role": "Co-founder and CEO",
      "company": "Superconductor",
      "twitter": "https://x.com/singharjun51293",
      "talks": [
        {
          "title": "Multiplayer agentic engineering: enabling your whole team and your best agents to work together",
          "description": "For a solo developer, coding agents are a superpower. For a team, they surface new kinds of bottlenecks: coordination, visibility, review, and shared context.\n\nWe wanted our whole team and our best agents to work together, with no work or context trapped on any one developer's machine. So we pressed pause on the product we were building to create a multiplayer cloud workspace for agentic engineering.\n\nThis talk shares five key practices we've learned from building and using our platform:\n\nTurn every surface the team uses into an agent interface.\nKick off sessions from Slack, review via iOS app, iterate in GitHub comments, ship from web. Agents run in the cloud, so work keeps moving even when your laptop is closed.\n\nMake agent work visible and collaborative across the whole team.\nEvery agent session is shared, has a live app preview, and an agent-guided code review. This allows engineers, PMs, and designers to steer and evaluate agent work collaboratively.\n\nTurn every external signal into shipped code your team can quickly evaluate.\nAutomatically turn customer emails, meeting action items, and bug reports into agent implementations that the whole team can review.\n\nSet up shared cloud dev environments so agents aren't siloed to individual machines.\nSecrets, role-based access, and network controls shared across the whole team. Fast environment startup, so you're not giving up speed by moving off local.\n\nBenchmark agents on your own codebase.\nClaude Code, Codex, Gemini, Amp, OpenCode — how do you know which is actually better on your stack? We'll cover using your merged PRs as ground truth to build a \"Personal SWE-Bench\" for your codebase.\n\nAgentic engineering is going multiplayer. This is how your team gets there.",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.009470983,
        0.06126992,
        0.025832968,
        0.0184092,
        -0.02170891,
        -0.059666883,
        -0.082974255,
        0.014608555,
        0.011096345,
        -0.23874043,
        -0.05447104,
        0.0014264039,
        0.012655303,
        -0.07531402,
        -0.04073978,
        -0.01742982,
        0.1201979,
        -0.10578688,
        -0.079676405,
        -0.048592728,
        0.0110713225,
        -0.07809476,
        -0.000019168254,
        0.027302017,
        -0.008383642,
        0.15637538,
        -0.024792753,
        0.060458947,
        -0.12029507,
        0.42409277,
        -0.02557047,
        -0.0033955823,
        0.033532187,
        -0.041642003,
        0.066199444,
        -0.06765966,
        -0.009255835,
        -0.016987745,
        0.019415224,
        0.106026545,
        -0.021664511,
        0.100067355,
        -0.044748943,
        0.0378818,
        -0.039440528,
        -0.057602137,
        -0.078722574,
        -0.0130214635,
        0.009116359,
        -0.013405357,
        -0.07906276,
        0.00067199103,
        0.037960768,
        -0.0012076241,
        0.06608824,
        0.07466992,
        -0.018303305,
        -0.08556285,
        0.07599333,
        -0.059960127,
        -0.08799292,
        -0.0067983516,
        0.04910376,
        0.052163403,
        -0.04545188,
        -0.101499215,
        0.0474233,
        -0.036553875,
        0.030816501,
        -0.0861864,
        0.025334822,
        0.028244277,
        -0.03219979,
        0.040450335,
        -0.0116266925,
        -0.14910412,
        0.00094192964,
        0.06292457,
        -0.04673213,
        0.04084474,
        -0.09527405,
        0.00869228,
        -0.06934782,
        -0.055340704,
        0.05957554,
        -0.0077696308,
        -0.056147426,
        0.010962151,
        0.0037770937,
        0.060911678,
        -0.030925589,
        0.023874065,
        -0.040853623,
        0.008277546,
        -0.051180426,
        -0.0066630132,
        -0.04310391,
        0.00066942256,
        -0.019162025,
        0.029471168,
        0.056145784,
        -0.07081496,
        0.041456062,
        0.007766317,
        -0.01396723,
        -0.014778142,
        0.046573456,
        -0.051086586,
        -0.011363572,
        -0.038106453,
        0.026738003,
        -0.6097603,
        0.052571665,
        0.03137088,
        -0.050574113,
        -0.14659163,
        0.048269387,
        -0.02252787,
        0.03720218,
        0.05078915,
        -0.03901004,
        0.14864911,
        -0.059198838,
        0.021981936,
        -0.040103406,
        -0.0017537331,
        0.07582365,
        0.011648903
      ]
    },
    {
      "id": "worldsfair-speaker-328",
      "conference": "worldsfair",
      "name": "James Le",
      "role": "Head of Developer Experience",
      "company": "TwelveLabs",
      "twitter": "https://x.com/le_james94",
      "talks": [
        {
          "title": "Video Has No Memory. Here's How We Built One.",
          "description": "Every video AI query today starts from scratch. There's no durable state, no entity continuity, no way to ask \"what does this corpus know?\" instead of \"find me something like this.\" This talk is about fixing that by engineering a proper memory layer for video intelligence, grounded in what we shipped at TwelveLabs with Jockey. What this talk covers: 1 - Why video memory is categorically different from text memory: Video is temporal, multimodal, dense, ambiguous, and evidence-sensitive. Larger context windows don't solve this. The problem isn't retrieval bandwidth, it's that there's no durable representation to retrieve into. 2 - The context graph as a systems concept, not a database choice: I'll define what \"context graph\" actually means in practice: time-bounded moments, cross-video entity resolution, appearance tracking, and relationship mapping. This is infrastructure-level thinking, not a graph DB sales pitch. 3 - Five design principles that determine whether video intelligence is reusable infrastructure or a search wrapper with extra steps: + Ingest once, reason many times (move expensive understanding work into preparation) + Store primitives, not just answers (moments, entities, appearances, relationships) + Ground every claim to source video (a timestamp is a product requirement, not a safety footnote) + Let intent shape memory (brand safety and sports highlights need different primitives from the same footage) + Keep the memory layer composable and API-first 4 - What this unlocks for builders. Corpus digest, agentic search with grounded references, entity-centric workflows, timeline reconstruction, and compliance tooling, all built on the same durable substrate. The talk is concrete and demo-grounded. You'll leave with a specific mental model for memory architecture, actionable decisions for ingestion pipeline design and entity resolution, and a clear line between \"search with extra steps\" and actual video intelligence infrastructure.",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.045139603,
        0.021673376,
        0.0025422324,
        0.017534388,
        0.005030013,
        -0.02705147,
        -0.036471415,
        0.09799333,
        0.053874034,
        -0.25242546,
        -0.040435474,
        -0.05273174,
        -0.036347866,
        -0.009051231,
        -0.030196,
        0.027129702,
        0.07125135,
        -0.08802603,
        -0.08024321,
        -0.07103905,
        -0.07180963,
        -0.027335279,
        -0.08561224,
        0.08596966,
        -0.028835159,
        0.11206098,
        0.023176268,
        0.05686747,
        -0.12850648,
        0.42378536,
        -0.07700371,
        0.028018115,
        -0.059554223,
        -0.055213902,
        0.029836688,
        -0.06592967,
        -0.0051636985,
        0.030090302,
        0.0017554834,
        0.03695685,
        -0.03095718,
        0.027865693,
        0.008448065,
        0.02828851,
        0.008425446,
        -0.013058577,
        -0.03803183,
        0.05658866,
        0.030027254,
        0.014283429,
        -0.0656536,
        -0.035282813,
        0.06995995,
        0.02715974,
        0.0031410363,
        0.048708778,
        0.03784234,
        0.0059844158,
        0.10664336,
        -0.05327355,
        -0.10947078,
        -0.053281076,
        0.039670017,
        -0.017779287,
        -0.038259156,
        -0.06818231,
        0.062173013,
        -0.031159401,
        -0.018938422,
        -0.042553406,
        -0.04320639,
        0.015128332,
        -0.04684217,
        0.03121803,
        -0.1226558,
        -0.11157967,
        -0.014395062,
        0.09722406,
        -0.018866584,
        0.029051209,
        0.038100366,
        0.08199531,
        -0.027369488,
        -0.0418194,
        0.04391618,
        -0.024870938,
        -0.10090677,
        0.05323074,
        -0.039151248,
        0.094124936,
        -0.031033298,
        -0.038825516,
        -0.055697102,
        -0.0154850045,
        -0.09421869,
        0.03777559,
        -0.015115001,
        0.050355237,
        0.010554453,
        0.019620586,
        -0.031967368,
        -0.03936864,
        -0.05713413,
        0.034730457,
        0.006128023,
        -0.07184584,
        0.015265574,
        -0.023702156,
        0.10299595,
        0.013624413,
        -0.047304384,
        -0.6231346,
        0.013095558,
        -0.030191192,
        0.00022535247,
        -0.14452688,
        0.033609178,
        -0.03535716,
        -0.0037815743,
        0.05355516,
        -0.025286324,
        0.1231843,
        0.02127681,
        0.008812064,
        -0.0067789205,
        0.0436026,
        0.06718128,
        -0.025339508
      ]
    },
    {
      "id": "worldsfair-speaker-329",
      "conference": "worldsfair",
      "name": "Christopher Burns",
      "role": "Founder & CEO",
      "company": "Inth",
      "twitter": "https://x.com/burnedchris",
      "talks": [
        {
          "title": "How We Got LLMs to Recommend Our Open Source Library (Without Paying or Plug-ins)",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.03534929,
        0.0119912615,
        -0.014087816,
        0.028779332,
        0.025778161,
        -0.05835212,
        -0.061637457,
        -0.0062981662,
        0.02937136,
        -0.24308458,
        -0.021573534,
        -0.007918128,
        0.0033632892,
        -0.09716566,
        -0.06812725,
        0.044665154,
        0.09627113,
        -0.1332595,
        -0.06619549,
        -0.10301869,
        -0.03873042,
        -0.100255996,
        -0.009723014,
        0.007692313,
        -0.076642334,
        0.063861534,
        -0.020610338,
        0.050217405,
        -0.11513225,
        0.43241215,
        -0.08240359,
        0.020440951,
        -0.017123925,
        0.002427068,
        0.08469213,
        -0.017976463,
        0.007796938,
        0.03809277,
        0.0010430965,
        0.038733535,
        -0.015373811,
        -0.0062997737,
        0.032611564,
        -0.039391503,
        0.041768808,
        -0.044003498,
        -0.046425574,
        -0.06843812,
        -0.015479512,
        -0.018413195,
        -0.06598498,
        -0.009094346,
        -0.061724745,
        -0.009318601,
        0.08519792,
        0.031623155,
        -0.0040865876,
        -0.056312755,
        0.13796084,
        0.045381773,
        -0.13302658,
        -0.048699126,
        -0.013994743,
        -0.0022036068,
        0.01644686,
        0.004895645,
        0.1297907,
        -0.044604685,
        -0.035429314,
        -0.06323004,
        -0.03521143,
        0.021146422,
        -0.078715,
        0.019835787,
        -0.07612495,
        -0.07358004,
        0.02530143,
        0.05749569,
        -0.0011466758,
        -0.0011480615,
        0.07305599,
        0.11611208,
        0.018379,
        -0.073949024,
        0.0454755,
        -0.035460103,
        -0.027211372,
        0.00076818716,
        -0.044216987,
        0.07037504,
        0.02396012,
        -0.07030744,
        -0.029916458,
        -0.061593715,
        -0.047544073,
        0.0051671467,
        -0.06293973,
        0.067525655,
        0.015337169,
        -0.059569307,
        -0.07155186,
        -0.006000856,
        0.0052758865,
        -0.040775724,
        -0.010402984,
        -0.006239827,
        0.014225444,
        -0.03144449,
        0.016329369,
        0.0054181325,
        -0.03826531,
        -0.6209554,
        0.004286851,
        -0.033221323,
        -0.047114134,
        -0.14303415,
        0.0160458,
        -0.006231937,
        -0.045949843,
        0.039098494,
        0.014731357,
        0.08597396,
        -0.036798853,
        0.06520543,
        0.04655006,
        0.07192542,
        0.021088636,
        -0.024749381
      ]
    },
    {
      "id": "worldsfair-speaker-330",
      "conference": "worldsfair",
      "name": "Vasant Kearney",
      "role": "CEO",
      "company": "Onlay AI",
      "twitter": "https://x.com/vasantkearney",
      "talks": [
        {
          "title": "Healthcare’s Agent Bytecode: X12 as the Harness for AI Agents",
          "description": "LLMs made old languages newly useful: COBOL for mainframes, Fortran for scientific code, and Rust, SQL, and Prolog as strict substrates for agentic systems. Healthcare has its own old language hiding in plain sight: X12. Before LLMs, X12 was mostly treated as ugly plumbing: loops, delimiters, companion guides, clearinghouse edits, payer-specific quirks, rejections, and acknowledgments. In an agentic workflow, those constraints become the feature. They give stochastic agents a deterministic target. This talk shows how healthcare agents can compile messy operational evidence into X12-shaped workflows: chairside audio into 837D claim narratives, imaging systems into 275/PWK attachment flows, payer portals and phone calls into 270/271 eligibility and 276/277 claim status, preauth evidence into 278 workflows, and EOBs, scanned mail, and bank data into 835/820 payment reconciliation. The core pattern is simple: LLMs reason over ambiguity; X12 provides the syntactic and semantic harness for validation, auditability, acknowledgments, rejections, human review, and high-volume automation. This is not an EDI nostalgia talk. It is a production architecture talk about building reliable agents in one of the messiest enterprise domains.",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05277952,
        0.047342043,
        0.059041604,
        0.022099724,
        0.023181139,
        0.025978928,
        -0.04998176,
        -0.0055439835,
        0.092040315,
        -0.2089196,
        -0.02857125,
        -0.0799332,
        0.009596125,
        -0.057037894,
        -0.07404601,
        0.031637702,
        0.1080905,
        -0.084080875,
        -0.09301995,
        -0.07625041,
        -0.021851186,
        -0.08893049,
        -0.020722361,
        0.08995568,
        -0.08593551,
        0.10993733,
        -0.05023098,
        0.07976148,
        -0.06787189,
        0.43096972,
        -0.06511923,
        -0.026353147,
        0.060252085,
        -0.028833034,
        -0.0012692465,
        -0.11006525,
        0.032904353,
        -0.07450749,
        -0.012406708,
        -0.010080825,
        0.011036248,
        0.05440711,
        0.004034181,
        0.019095717,
        -0.03453811,
        -0.025469039,
        -0.021645937,
        -0.01201465,
        -0.017837126,
        -0.09463468,
        -0.08169462,
        -0.0061121373,
        0.043195613,
        0.025901807,
        0.04538505,
        0.015891192,
        0.018736701,
        -0.04008695,
        0.062576815,
        -0.019393357,
        -0.08423257,
        -0.027796669,
        -0.0006529651,
        0.06018094,
        -0.0006540038,
        -0.09735088,
        0.064846955,
        0.032224264,
        0.016305108,
        -0.007837412,
        -0.0230019,
        -0.0013122029,
        -0.081103615,
        0.05347603,
        -0.09538228,
        -0.07455926,
        -0.10808087,
        0.07663018,
        0.009449978,
        0.03320736,
        0.015010755,
        0.07204847,
        -0.05709185,
        -0.024009632,
        0.009355328,
        -0.07882418,
        -0.0668109,
        0.0414236,
        -0.009097836,
        0.07687433,
        0.07177724,
        -0.07774561,
        -0.06638394,
        -0.020897396,
        -0.029436305,
        0.06801133,
        -0.071736455,
        0.012382088,
        -0.065070935,
        -0.00185642,
        -0.002052236,
        -0.04541217,
        0.080128685,
        -0.0287578,
        0.023866713,
        -0.019316565,
        0.02182727,
        -0.060330205,
        0.057548106,
        -0.059599143,
        -0.042426143,
        -0.6155717,
        0.0132405795,
        -0.018787064,
        0.012509083,
        -0.14312272,
        0.070290394,
        -0.010198858,
        0.03666521,
        0.06214667,
        -0.03906335,
        0.06769053,
        -0.010405635,
        -0.041903205,
        -0.020889023,
        -0.0356994,
        0.12687814,
        -0.0022438115
      ]
    },
    {
      "id": "worldsfair-speaker-331",
      "conference": "worldsfair",
      "name": "Gavin Uberti",
      "role": "Co-Founder & CEO",
      "company": "Etched",
      "twitter": "",
      "talks": [
        {
          "title": "Gavin Uberti — transformer-only ASICs for inference",
          "description": "Etched's Sohu approach to transformer inference on custom silicon.",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.01569418,
        0.06398036,
        0.034515254,
        0.047210597,
        0.03881779,
        0.014211104,
        -0.089768544,
        0.027475765,
        0.024950951,
        -0.18656011,
        0.040576942,
        0.06911796,
        -0.0061055,
        -0.0034941542,
        -0.100082815,
        0.02907167,
        0.028445104,
        -0.00670131,
        -0.032472502,
        -0.07186553,
        0.005627149,
        -0.082426675,
        -0.02746335,
        0.008518226,
        -0.04299759,
        0.11648787,
        0.0782349,
        0.07411151,
        -0.14525682,
        0.4543645,
        -0.09995586,
        -0.01879338,
        0.03959184,
        -0.017512836,
        0.071600385,
        -0.0602792,
        -0.016607182,
        -0.03718063,
        0.036370654,
        0.0140701905,
        0.045607783,
        -0.00051609834,
        -0.03907192,
        0.015671493,
        -0.07415259,
        -0.04652461,
        -0.018306319,
        0.0016221746,
        -0.018272972,
        0.017783979,
        -0.07365381,
        0.019680988,
        -0.0016974804,
        0.002270194,
        -0.011766754,
        0.019361148,
        -0.002957581,
        -0.07426406,
        0.08269886,
        -0.029840922,
        -0.13344924,
        -0.055009503,
        0.0032530308,
        0.0012406061,
        0.014154083,
        -0.08069409,
        0.103797205,
        -0.044063136,
        -0.07523463,
        -0.040349264,
        -0.0064874133,
        0.07991596,
        -0.077342704,
        -0.017462896,
        -0.05725295,
        -0.07283404,
        -0.07874433,
        0.09295888,
        -0.0005509535,
        -0.061980844,
        0.06372193,
        0.014073901,
        -0.045347046,
        -0.0016586815,
        0.020697702,
        -0.042953935,
        -0.015186791,
        0.019681193,
        -0.004472643,
        0.052380104,
        0.0078014494,
        0.007438437,
        -0.04891766,
        -0.06759209,
        0.0075925197,
        0.07351576,
        -0.04160545,
        -0.0008904293,
        0.0040011713,
        -0.017408745,
        0.010158013,
        0.02045248,
        0.01707241,
        -0.017764645,
        -0.066279545,
        -0.0364586,
        0.032694336,
        0.0041164733,
        -0.0046023005,
        -0.016377138,
        -0.03428673,
        -0.6492919,
        -0.01746582,
        0.0433666,
        -0.078100085,
        -0.15482365,
        -0.025487069,
        -0.05267269,
        0.066707425,
        0.029815165,
        0.0005760784,
        0.00838159,
        0.0068251765,
        0.09279155,
        0.017999345,
        -0.08149034,
        0.028201256,
        -0.028421555
      ]
    },
    {
      "id": "worldsfair-speaker-332",
      "conference": "worldsfair",
      "name": "Brian Lewis",
      "role": "AI Product Lead",
      "company": "Millennium",
      "twitter": "",
      "talks": [
        {
          "title": "Which AI startups actually land enterprise contracts? Lessons from evaluating 100+ AI startups at Millennium Management",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.014058523,
        0.025314964,
        -0.07837692,
        0.03210078,
        -0.06217111,
        -0.060883567,
        -0.02942457,
        0.025115008,
        0.026279336,
        -0.22042264,
        -0.00056194566,
        -0.06595204,
        -0.016378822,
        -0.055179052,
        -0.026117036,
        0.060464162,
        0.05272649,
        0.014247685,
        -0.049045943,
        -0.06449805,
        -0.03284217,
        -0.04826742,
        -0.016011419,
        0.03543889,
        -0.08883424,
        0.094152525,
        -0.055398095,
        0.061469283,
        -0.12741616,
        0.49624002,
        -0.06418116,
        0.0139850285,
        0.025622787,
        0.061920267,
        0.022896145,
        -0.049339853,
        -0.010473289,
        -0.034519047,
        0.035056908,
        0.03178675,
        -0.0047594267,
        0.04483038,
        -0.0118251,
        0.033191275,
        0.0033274922,
        0.01295447,
        -0.063503556,
        -0.05273176,
        0.041653484,
        -0.058666073,
        -0.03772548,
        -0.017764147,
        0.03347569,
        -0.019887032,
        0.040580317,
        0.00804848,
        0.0032820601,
        0.010765322,
        0.02069129,
        -0.012788445,
        -0.103424825,
        0.046857256,
        0.04421646,
        -0.014255302,
        0.047983836,
        -0.039336827,
        0.056955032,
        -0.021824233,
        0.007060648,
        -0.013191172,
        -0.10313631,
        0.054202624,
        -0.0659555,
        0.026465027,
        -0.07478263,
        -0.08765304,
        -0.0014887198,
        0.07887112,
        -0.0345836,
        0.025512423,
        0.05725475,
        0.011226765,
        -0.021639168,
        -0.085727446,
        -0.005618604,
        -0.016451778,
        -0.04811614,
        0.025452819,
        0.022583755,
        0.04403502,
        -0.023485288,
        -0.050438337,
        -0.04159737,
        -0.03792764,
        -0.08909532,
        0.06968565,
        -0.013678905,
        0.015791755,
        -0.012198043,
        -0.0060713827,
        -0.043279327,
        -0.03665646,
        0.057958327,
        0.041544005,
        0.0050498443,
        -0.00521124,
        -0.00680505,
        -0.016825981,
        0.064622335,
        -0.079831295,
        -0.08850413,
        -0.6442343,
        -0.02388581,
        -0.024614265,
        -0.037696354,
        -0.14401825,
        0.04766319,
        -0.027635364,
        0.01218325,
        0.09877143,
        0.038250994,
        -0.02522047,
        -0.04070964,
        -0.025045624,
        0.011663762,
        0.023638384,
        0.02662458,
        -0.03234301
      ]
    },
    {
      "id": "worldsfair-speaker-333",
      "conference": "worldsfair",
      "name": "Zach Bratun-Glennon",
      "role": "General Partner",
      "company": "Gradient",
      "twitter": "https://x.com/thezbg",
      "talks": [
        {
          "title": "Why your LLM is slow and expensive: lessons learned from running models in production",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "1:55pm-2:15pm",
          "type": "session"
        },
        {
          "title": "Scaling AI systems: where theory meets constraint",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.008829235,
        0.084022306,
        0.049131386,
        0.019989057,
        0.043590885,
        -0.039523628,
        -0.049729176,
        -0.0102837505,
        0.117007464,
        -0.21089575,
        -0.067090645,
        -0.050798334,
        0.011019696,
        -0.035649955,
        -0.061667453,
        0.0022684091,
        0.07960544,
        -0.06608047,
        -0.090078235,
        -0.059188038,
        -0.003013803,
        -0.12916636,
        -0.051196214,
        -0.004777516,
        -0.05613581,
        0.07488213,
        0.02200399,
        0.012481346,
        -0.10684315,
        0.48053107,
        -0.08762035,
        0.0040001613,
        0.03582194,
        0.038754594,
        0.03559881,
        -0.060268395,
        0.008551801,
        -0.027065966,
        0.038368262,
        0.067018025,
        0.020090869,
        0.07121482,
        -0.019430831,
        -0.057018034,
        -0.03862012,
        -0.003784715,
        -0.069236636,
        0.011727665,
        -0.04003277,
        -0.024859378,
        -0.050296996,
        0.0717595,
        0.08928448,
        -0.04105441,
        0.010403277,
        0.015620621,
        -0.06048815,
        -0.05970088,
        0.097125985,
        -0.07969532,
        -0.14676856,
        -0.012941727,
        -0.03459277,
        0.04525164,
        -0.014659448,
        0.025533505,
        0.07786363,
        0.028831074,
        0.033621117,
        -0.03206917,
        0.053981226,
        0.009146289,
        -0.03018483,
        0.054257907,
        -0.056157753,
        -0.09571485,
        0.009942488,
        0.08876945,
        -0.034062557,
        -0.006690033,
        0.036497876,
        0.030139571,
        -0.012374873,
        -0.00034522393,
        -0.0011958441,
        -0.03630273,
        -0.063907996,
        0.113582246,
        0.012917487,
        0.064831026,
        0.06870998,
        -0.037115257,
        -0.034637406,
        -0.016539415,
        0.010243251,
        0.08287354,
        -0.03228159,
        0.03151427,
        0.080210574,
        -0.06865603,
        0.008087866,
        0.0014441286,
        0.027180895,
        -0.0011787782,
        -0.05669487,
        -0.068068355,
        0.004013838,
        0.08525375,
        0.014810834,
        -0.023672553,
        -0.062302917,
        -0.5855631,
        -0.025669336,
        -0.005782495,
        -0.057025187,
        -0.18626054,
        0.06439361,
        0.017551156,
        -0.018938854,
        0.0718015,
        0.0396867,
        -0.004256194,
        0.015526614,
        0.04505487,
        0.08080927,
        0.022668857,
        0.015105063,
        -0.01799962
      ]
    },
    {
      "id": "worldsfair-speaker-334",
      "conference": "worldsfair",
      "name": "Sarah Sanders",
      "role": "",
      "company": "PostHog",
      "twitter": "",
      "talks": [
        {
          "title": "We let an AI agent execute Bash and lived to talk about it",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.026998186,
        -0.016597807,
        0.10542193,
        0.016010685,
        0.019483205,
        0.04156619,
        -0.07738423,
        -0.035024058,
        0.0646428,
        -0.2065821,
        -0.014200271,
        -0.04752887,
        0.016529063,
        -0.014343416,
        -0.083757296,
        0.030934518,
        0.096358396,
        -0.060119234,
        -0.04088624,
        -0.07711563,
        0.055418834,
        -0.058138896,
        0.0025034714,
        -0.002465462,
        -0.095356345,
        0.08642675,
        -0.023466634,
        0.045184836,
        -0.10190736,
        0.47920343,
        -0.10247472,
        0.038001824,
        0.009315043,
        -0.04916218,
        0.100684606,
        -0.067556,
        0.00180285,
        0.0078036105,
        -0.015720867,
        0.04811944,
        -0.0044412245,
        -0.015281006,
        -0.070279494,
        0.051246617,
        -0.1288872,
        0.026642224,
        -0.041656226,
        -0.068871155,
        -0.016598957,
        -0.0072700945,
        -0.08333719,
        -0.01945709,
        0.0018873975,
        -0.092473134,
        0.0263814,
        0.005548387,
        -0.027003445,
        -0.083021395,
        0.11343209,
        0.03216089,
        -0.08270388,
        -0.014312472,
        -0.059023727,
        0.0052531036,
        0.03688255,
        -0.044233546,
        0.113825664,
        -0.014797779,
        -0.0073426426,
        -0.04857065,
        0.024354458,
        0.020839447,
        -0.053385887,
        0.0042332066,
        -0.1314173,
        -0.07945098,
        0.0030045155,
        0.026125079,
        -0.0060858023,
        0.01881316,
        0.009345152,
        0.019344036,
        -0.1137848,
        -0.04789567,
        0.01377012,
        -0.01843161,
        -0.062051218,
        0.03462573,
        -0.028626427,
        0.013703001,
        0.026128057,
        -0.0033652668,
        -0.0025756254,
        0.07639193,
        -0.07633743,
        0.026239516,
        -0.044303905,
        0.025650859,
        -0.027859475,
        -0.010207917,
        -0.06536019,
        -0.08485246,
        0.012566377,
        -0.04860503,
        -0.013555607,
        -0.013057339,
        0.008523567,
        -0.07114731,
        0.0552305,
        -0.0040667416,
        -0.06842101,
        -0.58141243,
        -0.06250705,
        -0.05574778,
        -0.044727936,
        -0.14621073,
        -0.019512258,
        0.06582192,
        0.04277188,
        0.12628281,
        -0.04508133,
        0.08188947,
        -0.06468957,
        -0.03336462,
        -0.010801496,
        0.026284743,
        0.049696643,
        0.014015941
      ]
    },
    {
      "id": "worldsfair-speaker-335",
      "conference": "worldsfair",
      "name": "Keegan McCallum",
      "role": "",
      "company": "uRun",
      "twitter": "https://x.com/keeganmccallum3",
      "talks": [
        {
          "title": "Generative Video at the Speed of Light",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.019006155,
        0.07568013,
        0.08292982,
        0.07096411,
        -0.015794387,
        -0.03588216,
        -0.0006606063,
        0.0015954985,
        0.03291988,
        -0.2194151,
        -0.016241584,
        -0.0032573945,
        0.0013209515,
        0.04602121,
        -0.11056837,
        0.063341014,
        0.128655,
        -0.06237859,
        0.029991023,
        -0.15588968,
        0.051776342,
        -0.14735956,
        -0.010286753,
        -0.061987333,
        -0.073658094,
        0.07224356,
        -0.0008036865,
        -0.020635396,
        -0.064169005,
        0.4669556,
        -0.009617506,
        0.018883439,
        -0.020913525,
        0.013578634,
        0.070918836,
        -0.050116207,
        -0.016245248,
        -0.011816941,
        -0.025383765,
        0.049432006,
        -0.060020912,
        0.02499976,
        -0.08926887,
        0.014078951,
        -0.007003511,
        0.040036574,
        -0.06039863,
        -0.094818205,
        0.031599276,
        -0.04070186,
        -0.025877485,
        -0.026036263,
        -0.032249242,
        -0.07775161,
        0.026315676,
        -0.031023996,
        -0.023839321,
        -0.009355774,
        0.12223475,
        -0.047054425,
        -0.015326453,
        0.043036852,
        -0.001930508,
        -0.0008668922,
        -0.021546528,
        -0.07651303,
        0.11730777,
        0.007984659,
        -0.05550644,
        -0.023337174,
        -0.011974053,
        -0.037020385,
        -0.052499246,
        0.016443882,
        -0.053077262,
        -0.0019427242,
        -0.0630613,
        0.05289197,
        -0.014984904,
        0.0068671387,
        0.012881159,
        -0.021230973,
        0.0033121975,
        -0.027025985,
        -0.0522641,
        -0.08319851,
        -0.107996866,
        0.024922127,
        -0.03561825,
        0.074214436,
        0.005691394,
        0.015617672,
        -0.043034796,
        -0.033850092,
        0.023267118,
        0.06427133,
        -0.024122411,
        0.01760176,
        0.027678873,
        -0.046011917,
        -0.101452485,
        -0.03940675,
        -0.025804348,
        -0.023572661,
        -0.036475446,
        -0.06706067,
        -0.009496634,
        -0.02827554,
        0.012464877,
        -0.053194012,
        -0.03073234,
        -0.6173539,
        -0.042810313,
        -0.030649377,
        -0.03705929,
        -0.15314955,
        -0.07657347,
        -0.056566678,
        0.010971188,
        0.030584887,
        -0.014578128,
        0.040212486,
        0.037071113,
        0.06521415,
        0.0026760863,
        -0.03189615,
        0.047726914,
        -0.025548443
      ]
    },
    {
      "id": "worldsfair-speaker-336",
      "conference": "worldsfair",
      "name": "Ahnaf Prio",
      "role": "Senior Engineering Manager",
      "company": "Best Buy",
      "twitter": "",
      "talks": [
        {
          "title": "The Agentic Commerce Stack",
          "description": "Agents are already handling product discovery, cart building, and checkout — no human clicking required. But what's the protocol stack actually making this work? This talk maps the real infrastructure: MCP for tool access, A2A for agent coordination, the ACP spec (backed by OpenAI) and the UCP spec (backed by Google) — two competing approaches to standardizing the full agentic commerce lifecycle — and AP2 for agentic payments. We'll cover what each does, how they compose, and where they're still forming. Then we'll see it live — a working demo with a protocol inspector showing every tool call, task transition, and checkout event in real time. You'll leave with a clear mental model of the agentic commerce landscape and a reference implementation you can use.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.060577575,
        0.08041014,
        0.008534703,
        -0.020441564,
        0.029945645,
        -0.022595039,
        -0.083958164,
        -0.026719538,
        0.02115805,
        -0.20802298,
        0.01633459,
        -0.08653457,
        0.009827132,
        -0.07699923,
        -0.04433438,
        0.018093659,
        0.07914766,
        -0.07584787,
        -0.018474165,
        -0.03736246,
        -0.027182689,
        -0.101294994,
        -0.034642614,
        0.07781143,
        0.010149539,
        0.09855397,
        -0.009883131,
        0.02933036,
        -0.0693347,
        0.46699846,
        -0.081747524,
        0.006479543,
        -0.0018511725,
        -0.054520726,
        0.04133168,
        -0.087051615,
        -0.019060174,
        -0.0019522907,
        0.017651325,
        -0.0035333715,
        -0.020601954,
        0.06824147,
        -0.058919977,
        0.01138952,
        0.029469166,
        -0.018374365,
        -0.043248285,
        0.008134929,
        -0.027494889,
        -0.010095445,
        -0.074403696,
        -0.020432575,
        0.041118503,
        0.0076288856,
        0.04290687,
        -0.0010133104,
        0.0150755355,
        -0.048548628,
        0.060447656,
        -0.026488453,
        -0.07509026,
        0.014562224,
        0.031414803,
        0.029429782,
        -0.028109308,
        -0.12488932,
        0.07788679,
        -0.028194033,
        0.061663046,
        -0.081324734,
        -0.03117391,
        0.07027574,
        -0.08093918,
        0.0046938765,
        -0.022715438,
        -0.111139745,
        -0.08046447,
        0.10276939,
        -0.02088457,
        0.050649058,
        0.021245547,
        0.012588903,
        -0.09061162,
        -0.017050631,
        0.026240814,
        -0.0007536635,
        -0.015133859,
        0.02231612,
        0.0012314034,
        0.115899496,
        -0.016395966,
        -0.040200166,
        -0.09472801,
        0.07706603,
        -0.07218681,
        0.008170829,
        -0.03259578,
        0.08791627,
        0.004462495,
        0.056639895,
        -0.00625972,
        -0.058887035,
        0.09310808,
        0.018254155,
        0.0023911428,
        -0.013165749,
        0.014834279,
        -0.026766721,
        0.056297455,
        -0.047436107,
        -0.040042486,
        -0.61388224,
        -0.0017512531,
        0.020672077,
        0.023245435,
        -0.1519428,
        0.06045874,
        -0.025285674,
        0.05181585,
        0.1051962,
        -0.040691685,
        0.08823464,
        -0.038561184,
        -0.05722554,
        0.048811752,
        0.061433595,
        0.03009322,
        0.0143005615
      ]
    },
    {
      "id": "worldsfair-speaker-337",
      "conference": "worldsfair",
      "name": "Ramana Siddanth Emani",
      "role": "Data Scientist",
      "company": "Auditoria AI",
      "twitter": "https://x.com/siddanth2486",
      "talks": [
        {
          "title": "Your Finance Agent's Bottleneck Is You",
          "description": "Most \"AI for Finance\" demos look great and almost none survive past pilot. If you've pushed an agent past one workflow, one tenant, or one Workday schema, you know the bottleneck isn't the model - it's the engineer behind the agent, who can't iterate fast enough to keep up with real AP data, real RBAC, and real query volume. What if you built your dev loop with the same primitives you're shipping to the finance team? In this talk, I'll show the subagent + skills + MCP stack - a production multi-agent system over AP, PO, vendor, and multi ERP systems, a LangGraph pattern that survives production, and the three failure modes that kill finance pilots before they ship.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0018275524,
        0.040860917,
        -0.007026045,
        0.0015431762,
        0.06611621,
        -0.026519157,
        -0.09826393,
        -0.0044351565,
        0.05136806,
        -0.23308505,
        0.00013853282,
        0.011574447,
        0.024667224,
        -0.04613045,
        -0.06959245,
        0.028369155,
        0.084127896,
        -0.10087615,
        -0.07350928,
        -0.05430623,
        -0.055670094,
        -0.09680936,
        -0.048696402,
        0.05293597,
        -0.033144094,
        0.12800036,
        0.001046512,
        0.08193757,
        -0.11658092,
        0.46011883,
        -0.13599694,
        0.043586098,
        0.03331107,
        -0.021805963,
        0.08867939,
        -0.056904882,
        -0.0059526395,
        0.017206294,
        0.028388044,
        0.051070794,
        -0.026127413,
        0.05854345,
        -0.043591823,
        0.05619181,
        0.004743341,
        -0.026920825,
        -0.03676802,
        0.030267274,
        -0.051950164,
        -0.013120458,
        0.007947102,
        0.016938617,
        0.113894254,
        -0.02428518,
        0.062134802,
        0.04301463,
        0.038646426,
        -0.05941186,
        0.11580256,
        -0.051851224,
        -0.11172039,
        -0.027471267,
        0.010097772,
        -0.048153695,
        0.0084975725,
        -0.08402121,
        0.05448088,
        -0.028920153,
        0.014962383,
        -0.011763918,
        0.059925668,
        0.051822916,
        -0.07019916,
        0.04727541,
        -0.031935144,
        -0.15582627,
        -0.010294738,
        0.10315754,
        -0.00048138268,
        0.05066564,
        0.013892537,
        0.04877574,
        -0.06390578,
        -0.049405646,
        0.04181563,
        -0.02815976,
        -0.056597084,
        0.056260806,
        0.0077560074,
        0.06314035,
        -0.016701987,
        -0.04779689,
        -0.033980932,
        -0.09579253,
        -0.045746703,
        0.03436681,
        -0.061714146,
        0.04751869,
        -0.02689771,
        0.05052741,
        -0.015562358,
        -0.051176663,
        0.06849227,
        -0.02149897,
        -0.013008546,
        -0.05076926,
        0.046136234,
        -0.0125464825,
        0.023583842,
        0.03240254,
        -0.00096439855,
        -0.58896273,
        -0.0029699272,
        -0.02327317,
        -0.050149746,
        -0.14079209,
        0.06977059,
        -0.004309101,
        0.023603797,
        0.032592505,
        -0.07856491,
        0.06400306,
        0.010732063,
        -0.01943052,
        -0.0132132,
        0.025620604,
        0.05461372,
        -0.024215817
      ]
    },
    {
      "id": "worldsfair-speaker-338",
      "conference": "worldsfair",
      "name": "Shafik Quoraishee",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Local Agentic Theory For Accessible Mobile Games",
          "description": "Traditional mobile game architectures rely on static state machines and fixed behavioral trees. Under this model, gameplay and accessibility are treated as rigid, separate systems. This results in blunt difficulty toggles, predictable character loops, and reactive features that fail to address a player's actual context. Constraint-Centric Agentic Simulation (CCAS) offers a theoretical shift. By modeling the game world as a continuous, multi-agent negotiation, accessibility and challenge become part of a single, fluid continuum.\n\nUsing the JetBrains Koog framework on Android, this session explores the theory of running local agents on consumer mobile devices. We will discuss how principles of game theory, specifically dynamic negotiation and constraint satisfaction, can be used to build systems that reason over game states. Instead of executing pre-planned scripts, these agents dynamically alter their strategies. They negotiate environmental constraints to provide emergent challenges for high-skill players or organically smooth out cognitive and motor friction points for those requiring assistance.\n\nRunning these theoretical models on edge hardware requires overcoming significant practical hurdles. We will break down the architecture needed to support this continuous adaptation without relying on cloud computation. We will cover how to manage memory footprints, compress state histories for rapid backtracking, and schedule local planning loops so they integrate flawlessly with the rendering engine.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.05760963,
        0.069582544,
        0.0133753335,
        0.05462995,
        0.060990483,
        -0.02494437,
        -0.05430126,
        -0.05966429,
        0.08592064,
        -0.27376416,
        0.0037489412,
        -0.022212721,
        -0.040000066,
        -0.050370704,
        -0.007382716,
        -0.054265447,
        0.094040744,
        -0.031580444,
        -0.09210471,
        -0.045152247,
        -0.027131485,
        -0.09774402,
        -0.033168778,
        0.06619523,
        -0.07948646,
        0.116400115,
        -0.037253324,
        0.057189602,
        -0.076550916,
        0.46604195,
        -0.08253028,
        -0.0035293829,
        0.049226806,
        -0.0363946,
        -0.022672528,
        -0.017352084,
        -0.06822286,
        -0.0016866776,
        0.099057965,
        0.04736318,
        0.0075229607,
        0.0037396508,
        -0.067379944,
        0.061400346,
        0.028132543,
        -0.0018848008,
        -0.029796716,
        0.027884452,
        -0.021645516,
        0.019240312,
        -0.13072576,
        0.02995556,
        0.002471903,
        -0.051026054,
        0.03240618,
        0.05188869,
        -0.053120587,
        0.0033820262,
        0.0261797,
        0.03484643,
        -0.14367484,
        -0.00060729194,
        -0.05007155,
        0.01911009,
        0.046627566,
        -0.041538354,
        0.014409067,
        -0.02841957,
        -0.012437026,
        0.014209331,
        0.02677031,
        0.025248451,
        -0.015722869,
        -0.011155995,
        -0.08288452,
        -0.084840156,
        -0.072009414,
        0.101409405,
        -0.040751856,
        0.06454443,
        0.0023165534,
        0.034500614,
        -0.011252818,
        -0.038087904,
        0.047516715,
        -0.042622674,
        -0.018302396,
        0.08223548,
        -0.030689964,
        0.0626512,
        0.036008507,
        0.0062463665,
        -0.06836942,
        -0.005171691,
        -0.017157238,
        0.03567779,
        -0.065295696,
        0.008412723,
        0.0012105156,
        -0.0068212315,
        0.07163056,
        -0.08435663,
        0.044917695,
        0.013300194,
        0.047947362,
        -0.01799472,
        0.0015357803,
        0.008648922,
        0.028594417,
        -0.0042076316,
        -0.05798074,
        -0.59582555,
        -0.011831685,
        0.045369893,
        -0.06554211,
        -0.15444686,
        0.03730181,
        -0.036732912,
        0.061272167,
        0.07195989,
        -0.00035496196,
        0.096431084,
        0.012296915,
        -0.024860512,
        0.03350416,
        -0.04258961,
        0.077491194,
        0.035122618
      ]
    },
    {
      "id": "worldsfair-speaker-339",
      "conference": "worldsfair",
      "name": "Joanne Song",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Local Agentic Theory For Accessible Mobile Games",
          "description": "Traditional mobile game architectures rely on static state machines and fixed behavioral trees. Under this model, gameplay and accessibility are treated as rigid, separate systems. This results in blunt difficulty toggles, predictable character loops, and reactive features that fail to address a player's actual context. Constraint-Centric Agentic Simulation (CCAS) offers a theoretical shift. By modeling the game world as a continuous, multi-agent negotiation, accessibility and challenge become part of a single, fluid continuum.\n\nUsing the JetBrains Koog framework on Android, this session explores the theory of running local agents on consumer mobile devices. We will discuss how principles of game theory, specifically dynamic negotiation and constraint satisfaction, can be used to build systems that reason over game states. Instead of executing pre-planned scripts, these agents dynamically alter their strategies. They negotiate environmental constraints to provide emergent challenges for high-skill players or organically smooth out cognitive and motor friction points for those requiring assistance.\n\nRunning these theoretical models on edge hardware requires overcoming significant practical hurdles. We will break down the architecture needed to support this continuous adaptation without relying on cloud computation. We will cover how to manage memory footprints, compress state histories for rapid backtracking, and schedule local planning loops so they integrate flawlessly with the rendering engine.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.09421539,
        0.060716636,
        -0.0017114329,
        0.012938503,
        0.07611184,
        -0.051841427,
        -0.034323283,
        -0.048078302,
        0.06710539,
        -0.25432128,
        -0.05440359,
        0.029858138,
        -0.022577731,
        0.03147702,
        -0.029558066,
        0.0064405934,
        0.054361213,
        -0.027580995,
        -0.06681426,
        -0.07076891,
        -0.051319014,
        -0.06472839,
        -0.0026158672,
        0.034336783,
        -0.027385548,
        0.11710962,
        0.018577106,
        0.05873547,
        -0.07624206,
        0.4394478,
        -0.032697845,
        0.05766134,
        0.014327514,
        -0.03862946,
        0.034196157,
        -0.076531276,
        -0.05105756,
        -0.030297454,
        0.04937324,
        0.03631638,
        -0.0108354,
        -0.0042424235,
        -0.07568331,
        0.11442044,
        0.009935746,
        -0.024110349,
        -0.07244039,
        0.0070290514,
        -0.0020835968,
        0.031401925,
        -0.11613403,
        0.021491049,
        0.0038098164,
        -0.058192495,
        0.06335509,
        0.03101328,
        -0.0244423,
        -0.08593706,
        0.07939379,
        0.05983532,
        -0.13693081,
        0.026996262,
        -0.06586438,
        -0.022327827,
        0.025382875,
        -0.050955072,
        -0.030393148,
        -0.03314711,
        -0.041614138,
        -0.038566697,
        0.03596197,
        0.053741954,
        -0.0360479,
        -0.027311895,
        -0.07100616,
        -0.12792553,
        -0.010052483,
        0.08405906,
        -0.009130921,
        0.058199342,
        0.045542046,
        0.028319769,
        -0.01222722,
        -0.06383446,
        -0.010777171,
        -0.058230642,
        -0.056581046,
        0.101130284,
        -0.06156359,
        0.022757843,
        0.045069117,
        -0.007722538,
        -0.018229112,
        0.024555266,
        -0.10806085,
        -0.01264391,
        -0.054687604,
        0.02699557,
        -0.055147596,
        0.00065969635,
        -0.014656018,
        -0.04171634,
        -0.040388457,
        -0.027012827,
        0.0077421246,
        -0.019489609,
        -0.05296004,
        0.0069631627,
        0.015184528,
        0.0032603324,
        -0.05971235,
        -0.61057293,
        0.021525163,
        0.058982946,
        -0.029611802,
        -0.13970812,
        0.0008633977,
        -0.07471504,
        -0.000565641,
        0.12080233,
        -0.030328814,
        0.09651842,
        -0.0025992135,
        -0.015650516,
        0.005306674,
        -0.022496877,
        0.08395734,
        0.08433017
      ]
    },
    {
      "id": "worldsfair-speaker-340",
      "conference": "worldsfair",
      "name": "Jeff Wang",
      "role": "CTO",
      "company": "Exa",
      "twitter": "",
      "talks": [
        {
          "title": "Exa AI in GTM",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.026440794,
        0.092305735,
        -0.059696298,
        -0.024666453,
        0.043939825,
        0.0069866125,
        -0.077396974,
        0.005778984,
        0.038840365,
        -0.2102882,
        -0.06635735,
        -0.08860518,
        -0.034496028,
        -0.0055196504,
        -0.0072222003,
        0.074890174,
        0.016030567,
        -0.049162418,
        -0.07948894,
        -0.07592124,
        -0.046368085,
        -0.06714191,
        -0.003902634,
        0.03212457,
        -0.027151776,
        0.062081154,
        0.028787615,
        0.056940995,
        -0.106801935,
        0.46312737,
        -0.030487271,
        -0.025874263,
        0.0015823586,
        0.002615785,
        0.038908657,
        -0.065713316,
        -0.009434143,
        -0.03071298,
        0.051761486,
        0.03766821,
        0.011081177,
        0.06588279,
        -0.06592612,
        0.028949805,
        -0.008438425,
        -0.010390506,
        -0.08015625,
        0.0074158455,
        -0.023912981,
        -0.028585752,
        -0.07939007,
        0.021877918,
        -0.0007065364,
        -0.075313784,
        -0.026915764,
        -0.033705525,
        -0.0152378725,
        -0.049269613,
        0.022364527,
        -0.000006822311,
        -0.19530703,
        -0.022162504,
        0.0032672943,
        -0.01401216,
        0.0025774501,
        -0.040878516,
        -0.03519359,
        -0.007992352,
        -0.020438109,
        -0.039995883,
        0.01911728,
        -0.03183975,
        -0.035231967,
        0.014746334,
        -0.14950807,
        -0.049594965,
        0.005603973,
        0.08979767,
        -0.06779943,
        -0.030950392,
        0.050323173,
        0.08793651,
        -0.11744583,
        -0.073355526,
        0.006483156,
        -0.00012080596,
        -0.0741355,
        0.0015939733,
        -0.045713108,
        0.077130124,
        0.0225833,
        -0.0005560914,
        -0.05517237,
        0.027164381,
        -0.10262544,
        0.09565689,
        -0.044533387,
        0.077302754,
        -0.026645405,
        0.011445879,
        -0.058203444,
        0.06895171,
        0.045469787,
        0.0059416834,
        -0.07424222,
        -0.0735769,
        0.0066025327,
        -0.017791135,
        0.051830396,
        -0.019155204,
        -0.048497118,
        -0.60131884,
        0.022230037,
        -0.036125574,
        -0.031290762,
        -0.1265609,
        0.028052948,
        -0.060984787,
        0.099294536,
        0.07277429,
        -0.08430189,
        0.045235645,
        0.02417507,
        0.008930347,
        -0.022708373,
        0.09602385,
        -0.0355251,
        0.015942058
      ]
    },
    {
      "id": "worldsfair-speaker-341",
      "conference": "worldsfair",
      "name": "Ayush Bhardwaj",
      "role": "",
      "company": "Allos AI",
      "twitter": "https://x.com/aybh08",
      "talks": [
        {
          "title": "Trading Desks to Clinical Trials: Parallels in Applied Vertical AI",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.016982129,
        0.04478583,
        -0.0053829164,
        0.07499416,
        -0.0123172095,
        -0.032353677,
        -0.042614814,
        0.056132216,
        0.00038501254,
        -0.22046573,
        -0.07000456,
        0.016220676,
        0.017873142,
        -0.016836792,
        -0.0549746,
        0.051471163,
        0.08865327,
        -0.05658997,
        -0.029361397,
        -0.070977986,
        -0.03994025,
        -0.033017352,
        -0.07201369,
        0.018750073,
        -0.08152363,
        0.05258163,
        -0.025722079,
        0.049293265,
        -0.09787734,
        0.49658898,
        -0.06621036,
        -0.070634134,
        -0.035297856,
        0.0234474,
        0.065368,
        -0.0057019815,
        0.038007163,
        -0.007370886,
        0.046005316,
        0.038729884,
        -0.030713487,
        0.038227994,
        0.028892256,
        0.07433459,
        0.023021162,
        0.059790622,
        -0.058673285,
        -0.053852312,
        -0.029972496,
        -0.02869023,
        -0.010752088,
        0.0018752583,
        0.0071990285,
        -0.057735223,
        0.001165294,
        0.026660277,
        -0.040243696,
        -0.032723915,
        0.08268684,
        -0.03897966,
        -0.064295806,
        -0.026706327,
        0.022268245,
        0.016632563,
        0.022933586,
        -0.052416902,
        0.08037106,
        -0.10744694,
        -0.028123291,
        -0.010251943,
        -0.033430237,
        0.02110535,
        -0.062313225,
        0.000015699787,
        -0.0805976,
        -0.038333282,
        0.0029261126,
        0.14930852,
        0.014226063,
        0.06424102,
        -0.016812345,
        0.026771555,
        -0.061461974,
        -0.007203825,
        -0.031426795,
        -0.017190166,
        -0.04663393,
        0.098679624,
        -0.016909339,
        0.09754515,
        0.02089719,
        -0.026533479,
        -0.03776908,
        -0.025022304,
        -0.01627726,
        0.03384405,
        -0.036072623,
        0.021613406,
        0.025511032,
        -0.01541339,
        -0.0073892293,
        -0.024559887,
        0.06309298,
        0.022518607,
        -0.04987709,
        -0.04312691,
        0.036256477,
        0.0113900015,
        -0.020029902,
        0.011133764,
        -0.017495291,
        -0.6401931,
        -0.06397993,
        0.015860315,
        -0.04928848,
        -0.17284097,
        0.014077265,
        -0.006318159,
        0.043375224,
        0.060389213,
        -0.062513724,
        0.03896781,
        -0.021422753,
        0.0065614516,
        0.015308959,
        0.026258335,
        0.01358543,
        -0.03464498
      ]
    },
    {
      "id": "worldsfair-speaker-342",
      "conference": "worldsfair",
      "name": "Yuchen Fama",
      "role": "Software engineer",
      "company": "Red Hat / vLLM",
      "twitter": "",
      "talks": [
        {
          "title": "KV Cache-Aware Routing and P/D Disaggregation on Kubernetes: The Parts Public Benchmarks Don't Show",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.080095164,
        0.016342876,
        0.0072116503,
        0.016361753,
        0.06450494,
        0.0018257211,
        -0.03597196,
        0.043763693,
        0.05978073,
        -0.24019684,
        -0.021801978,
        -0.090291195,
        0.014526572,
        0.0041375114,
        -0.08211459,
        -0.007865059,
        0.07942972,
        -0.105384335,
        -0.04830005,
        -0.083863884,
        -0.01064583,
        -0.093027696,
        -0.012398848,
        -0.02678217,
        -0.094841525,
        0.043300997,
        0.017161278,
        0.029147983,
        -0.15898213,
        0.4681672,
        -0.09097903,
        -0.010374259,
        0.0038118553,
        -0.06780309,
        0.04889382,
        -0.028388878,
        -0.0013695218,
        -0.0072295903,
        0.062321898,
        0.066270255,
        0.019647531,
        0.030419605,
        -0.04366006,
        0.022909142,
        -0.14571434,
        -0.034589805,
        -0.09698696,
        0.025767215,
        -0.0324475,
        0.016511735,
        -0.10388213,
        -0.052246425,
        0.045282405,
        -0.011243321,
        -0.014546579,
        0.00022459135,
        -0.026845155,
        -0.04026905,
        0.09675887,
        0.0036766543,
        -0.12785992,
        0.021287022,
        0.006031968,
        -0.010355919,
        -0.0029527359,
        -0.032701597,
        0.034808844,
        -0.09506129,
        -0.09654095,
        0.037056673,
        -0.0028684274,
        -0.026549125,
        -0.030697603,
        0.0021344137,
        -0.0560794,
        -0.070560515,
        0.011217218,
        0.08927299,
        -0.013112503,
        0.02044877,
        0.044855632,
        0.10022307,
        -0.10341165,
        -0.02381996,
        0.041652214,
        0.010464608,
        -0.04682896,
        0.077478714,
        -0.03667772,
        0.054058913,
        0.047903072,
        -0.011744495,
        -0.031707574,
        -0.03266862,
        -0.055487007,
        0.07537782,
        0.005060608,
        0.09203807,
        -0.01626385,
        0.009862858,
        0.007856361,
        0.016344639,
        -0.02061343,
        -0.014282267,
        -0.02004114,
        -0.057339143,
        -0.01901296,
        -0.0062189302,
        0.027116993,
        -0.049537934,
        -0.049355824,
        -0.591635,
        -0.07118303,
        -0.015459799,
        -0.06137127,
        -0.14603648,
        0.025576772,
        -0.036804598,
        0.010000102,
        0.07040675,
        -0.05748993,
        0.028695824,
        0.020496162,
        0.04038475,
        0.023007656,
        0.05460072,
        0.031297326,
        0.008060274
      ]
    },
    {
      "id": "worldsfair-speaker-343",
      "conference": "worldsfair",
      "name": "Michey Mehta",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "KV Cache-Aware Routing and P/D Disaggregation on Kubernetes: The Parts Public Benchmarks Don't Show",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.018592183,
        0.12428871,
        -0.0019383464,
        0.03630736,
        0.051840067,
        -0.03077798,
        -0.09515723,
        0.0036286316,
        0.059003457,
        -0.21940073,
        -0.026905999,
        -0.029279044,
        -0.039356086,
        -0.03802523,
        -0.0649656,
        0.037056908,
        0.115193054,
        -0.04967398,
        -0.075065546,
        -0.101141736,
        -0.016318843,
        -0.09242177,
        -0.03358652,
        -0.0030901656,
        -0.059564658,
        -0.00019505598,
        -0.00090891705,
        0.018977208,
        -0.16228738,
        0.5009795,
        -0.09004343,
        -0.09338825,
        -0.028283415,
        -0.02914469,
        0.03263394,
        0.0024451127,
        0.013322028,
        0.019895652,
        -0.017747166,
        0.09464397,
        0.053551126,
        0.058734275,
        -0.012621497,
        -0.01860871,
        -0.08772167,
        -0.043037683,
        -0.091385856,
        0.00027834138,
        0.011621799,
        -0.04386917,
        -0.10142695,
        -0.023290332,
        -0.011685829,
        -0.03098872,
        -0.02284148,
        -0.001708019,
        -0.05716203,
        -0.08095442,
        0.06709247,
        -0.07611459,
        -0.05475399,
        0.04089484,
        -0.041259047,
        0.0041798134,
        -0.085503034,
        -0.04233025,
        0.08117801,
        -0.06913076,
        -0.03796065,
        0.034297395,
        0.046740822,
        -0.03398447,
        -0.058759864,
        0.04443439,
        0.0018744142,
        -0.058765963,
        0.0048509366,
        0.08775968,
        -0.038384892,
        -0.004815638,
        0.011169942,
        0.10215996,
        -0.027353056,
        0.010510783,
        0.009672151,
        -0.05855575,
        -0.06563219,
        0.029314354,
        -0.016687164,
        0.06921607,
        0.06575905,
        0.009568348,
        -0.053468734,
        0.015368343,
        -0.039983593,
        0.049214754,
        -0.016381683,
        0.1105896,
        -0.060454126,
        0.045514226,
        0.051540226,
        -0.010929046,
        0.008926031,
        -0.05098254,
        -0.053296603,
        -0.012356044,
        0.006413762,
        0.0018708423,
        0.01708598,
        -0.09609212,
        -0.059306443,
        -0.56786954,
        -0.0046125813,
        0.06056828,
        -0.07108143,
        -0.15950343,
        0.035667136,
        -0.0628911,
        -0.004359786,
        0.051019315,
        -0.024664054,
        -0.011851793,
        -0.03709767,
        -0.032055914,
        0.009241459,
        -0.018271577,
        0.035239056,
        0.012596121
      ]
    },
    {
      "id": "worldsfair-speaker-344",
      "conference": "worldsfair",
      "name": "Ashish Kamra",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "KV Cache-Aware Routing and P/D Disaggregation on Kubernetes: The Parts Public Benchmarks Don't Show",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.014652864,
        0.0956835,
        0.0072774063,
        0.05086908,
        0.083366886,
        -0.027003342,
        -0.092255,
        -0.0064615286,
        -0.0079585565,
        -0.24633998,
        -0.004989232,
        -0.022831773,
        0.0092165675,
        -0.05773302,
        -0.057996586,
        -0.05838165,
        0.045691043,
        -0.047974516,
        -0.043389596,
        -0.08814996,
        -0.002411504,
        -0.08562511,
        -0.04083088,
        -0.046910338,
        -0.04691475,
        0.05185073,
        0.0012177342,
        -0.015712388,
        -0.13306724,
        0.48805606,
        -0.04302759,
        -0.08711281,
        -0.012229534,
        -0.044879053,
        -0.017194998,
        0.022007674,
        0.019592784,
        0.034803465,
        0.04044432,
        0.08038798,
        -0.019640988,
        0.006413967,
        -0.025071494,
        0.03136645,
        -0.12416128,
        -0.04545095,
        -0.036352474,
        -0.024597405,
        -0.027609581,
        -0.031825013,
        -0.06159991,
        0.005165256,
        0.0035374693,
        -0.03551272,
        -0.013602869,
        0.021885559,
        -0.028480517,
        -0.032390412,
        0.058379985,
        0.0063424837,
        -0.12769951,
        0.01583261,
        -0.012602796,
        -0.010720965,
        -0.066903085,
        -0.037705246,
        0.08982418,
        -0.038303327,
        -0.11399199,
        0.05009747,
        0.052439157,
        -0.03426471,
        -0.04426238,
        -0.009304594,
        -0.014720996,
        -0.028684078,
        0.027583098,
        0.10955801,
        -0.061816968,
        0.028403549,
        -0.013684072,
        0.10071827,
        -0.066969104,
        -0.03622985,
        0.021934029,
        -0.013789506,
        -0.076786295,
        0.053387966,
        -0.03204378,
        0.079137124,
        0.077979386,
        -0.014049516,
        -0.053467162,
        -0.06257429,
        -0.036276404,
        0.022515556,
        -0.034914695,
        0.07282391,
        -0.0066723367,
        0.044600766,
        0.061740942,
        0.04260861,
        -0.030488476,
        -0.004011404,
        0.0042434987,
        0.009460856,
        0.006489956,
        -0.020250065,
        0.027215146,
        -0.11452273,
        -0.048706464,
        -0.6082904,
        -0.00008513896,
        0.014378044,
        -0.058484826,
        -0.1478116,
        -0.000060952727,
        -0.08045835,
        0.0055566323,
        0.021977842,
        -0.015134925,
        0.022929769,
        0.01429383,
        0.0037936696,
        0.046449173,
        0.009952093,
        0.00790559,
        -0.0066655916
      ]
    },
    {
      "id": "worldsfair-speaker-345",
      "conference": "worldsfair",
      "name": "Kunal Lanjewar",
      "role": "",
      "company": "Riot Games",
      "twitter": "",
      "talks": [
        {
          "title": "Your Hero Agent Needs a Party",
          "description": "A front-door persona, a party of deterministic specialist agents, A2A between. Your support bot deflects half its tickets, then, soloing a problem it was never built for, confidently runs the wrong `kubectl` command. Most teams respond by rewriting the prompt. The real fix is a multi‑agent party of specialists. This talk gives you a production pattern that turns one over-leveled hero agent into a coordinated party of specialists you can trust on tier-zero infrastructure. Persona and ReAct agents make great heroes at the front door. Any team can copy one, paste it into their stack, and adjust the behavior in plain English. But if you send a lone hero to clear the dungeon, whether it is a deploy or an incident, a non-deterministic Reason-Act loop tends to loop, over-act, or punt back to a human. More prompts and more skills do not reliably level it up. Instead of soloing, keep the persona as the front-door face and give it a party: deterministic DAG specialists where the graph is fixed and the LLM is called only at decision points. For example, a deployment specialist can list rolling pods, choose the next tool, run it, read logs, and then diagnose the result. Each specialist is a class with one job and a narrow set of tools, and they coordinate over A2A for capability discovery and delegation across frameworks. Reliability and tighter least-privilege access become properties of the design, not something you try to bolt onto a prompt. You’ll leave with the pattern: where to draw the line between the hero and its specialists, how to shape a DAG specialist so it decides instead of flails, and where A2A fits as the seam between them, grounded in lessons from a tier‑zero fleet.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.02517799,
        0.089986704,
        -0.0024585046,
        0.020988615,
        0.057285935,
        -0.05702615,
        -0.024150059,
        0.0012081722,
        0.04429026,
        -0.24085248,
        -0.027306456,
        0.039367568,
        0.004515082,
        -0.0027345212,
        -0.069298595,
        -0.0113896895,
        0.08108056,
        -0.08384225,
        -0.11406302,
        -0.06050452,
        -0.037762772,
        -0.08994861,
        -0.011717679,
        -0.0149050765,
        -0.0055495803,
        0.093656175,
        -0.010203613,
        0.014786888,
        -0.10959539,
        0.48007727,
        -0.046532687,
        0.008802875,
        0.029505923,
        -0.101152405,
        0.045183953,
        -0.009124122,
        -0.028758893,
        -0.014584918,
        0.01662641,
        -0.01936281,
        -0.0059682447,
        0.109207034,
        -0.06973366,
        0.056087673,
        -0.028569018,
        -0.017346647,
        -0.072111465,
        0.040123105,
        -0.024593351,
        -0.05194123,
        -0.033646915,
        0.037304413,
        -0.0016746207,
        -0.046544876,
        0.028224016,
        0.058950014,
        0.006989884,
        -0.054608267,
        0.079557136,
        0.017734319,
        -0.086698234,
        -0.054282103,
        -0.028405704,
        0.016854987,
        0.016894782,
        -0.060868148,
        0.021257257,
        -0.05330667,
        -0.029732738,
        -0.002755656,
        0.032855846,
        0.019760242,
        -0.024204198,
        0.030066421,
        -0.054879114,
        -0.057123866,
        -0.05721861,
        0.12610526,
        0.0009052024,
        0.014807882,
        0.037993684,
        0.045074876,
        -0.12244514,
        0.00009809348,
        0.05917579,
        -0.005716048,
        -0.07209983,
        0.0065358747,
        0.0075732674,
        0.11023748,
        0.010791113,
        -0.041842557,
        -0.09255644,
        -0.004784593,
        -0.037071586,
        -0.009198182,
        -0.08829085,
        0.018294433,
        -0.077003516,
        0.0694782,
        0.029183881,
        -0.065585546,
        -0.023217387,
        -0.011831973,
        0.0022076033,
        -0.057284743,
        -0.04777317,
        0.041812,
        0.017596127,
        -0.002689426,
        -0.016137458,
        -0.6077408,
        -0.012501481,
        -0.05542071,
        -0.013595081,
        -0.15175298,
        0.09362073,
        -0.023441153,
        0.030685864,
        0.09034226,
        -0.052285064,
        0.07515874,
        0.006285833,
        -0.0026397884,
        -0.0065785795,
        0.03435658,
        0.038521904,
        -0.005194263
      ]
    },
    {
      "id": "worldsfair-speaker-346",
      "conference": "worldsfair",
      "name": "Stephen Balaban",
      "role": "Co-founder / CTO",
      "company": "Lambda",
      "twitter": "https://x.com/stephenbalaban",
      "talks": [
        {
          "title": "Scaling AI systems: where theory meets constraint",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.05886767,
        0.047607265,
        0.06667016,
        0.054653037,
        -0.025941608,
        0.0014578735,
        -0.045977138,
        0.011921476,
        0.10348618,
        -0.2341208,
        -0.0009337654,
        -0.107281,
        0.0076402314,
        0.023951245,
        -0.08079368,
        0.03758462,
        0.14998733,
        -0.10224389,
        -0.045027368,
        -0.01639199,
        -0.02476021,
        -0.093207404,
        -0.028344868,
        0.002598697,
        -0.0689633,
        0.096362144,
        0.026233591,
        0.0068870466,
        -0.075685315,
        0.484171,
        -0.034006093,
        -0.03413481,
        -0.010171743,
        -0.017728262,
        0.08637724,
        -0.07706342,
        -0.0010527613,
        0.050140116,
        0.034439523,
        0.010596039,
        0.041654628,
        0.03228945,
        -0.030251209,
        -0.060099773,
        -0.05309339,
        0.0292215,
        -0.07226499,
        -0.0040087965,
        -0.04925651,
        0.009036789,
        -0.041475963,
        0.02786371,
        0.032943882,
        0.03669719,
        0.018212447,
        0.028750464,
        -0.018434567,
        -0.033216115,
        0.09236573,
        -0.023060147,
        -0.15073197,
        -0.042504106,
        -0.040503405,
        0.023444477,
        -0.0040488276,
        -0.072807714,
        0.11541637,
        -0.003611092,
        0.023665288,
        -0.05762407,
        0.05526504,
        0.025878165,
        -0.011861067,
        0.055789772,
        -0.08067651,
        -0.042596973,
        -0.010555673,
        0.025431229,
        0.012877267,
        0.03720364,
        0.036561057,
        0.008496652,
        -0.04307723,
        -0.06454342,
        0.027396329,
        -0.019394344,
        -0.051953852,
        0.055925325,
        0.032716412,
        0.037291575,
        0.02908005,
        -0.06309308,
        -0.019322036,
        -0.014560916,
        0.0128856255,
        0.0033986054,
        -0.09568332,
        -0.012514251,
        0.03962645,
        -0.0690044,
        0.020045348,
        -0.013135778,
        0.032061882,
        -0.03856617,
        -0.049309768,
        -0.07266178,
        0.04019079,
        0.06858208,
        0.066626884,
        -0.021728883,
        -0.0074730194,
        -0.60118115,
        -0.04329366,
        0.003664444,
        -0.04654254,
        -0.16466872,
        -0.017865345,
        -0.024335431,
        -0.011538494,
        0.11367776,
        -0.009643552,
        0.052501995,
        0.025496809,
        0.035208903,
        0.01789494,
        0.015611221,
        0.06214057,
        0.0022636852
      ]
    },
    {
      "id": "worldsfair-speaker-347",
      "conference": "worldsfair",
      "name": "Melanie Warrick",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Human Is an Async API Subtitle: Designing Durable Human-in-the-Loop Agents",
          "description": "Production agent systems need humans in the loop. So why do they keep getting modeled as synchronous tool calls? The agent ecosystem is focused on autonomy, but in reality, especially for high-stakes or regulated workflows, humans are a critical feature, not an afterthought. This demo-driven talk shows how to stop bolting on humans and start treating them as async-by-default endpoints with proper durability, retry, and escalation semantics. We will walk through two live, multi-agent patterns built with LangGraph and Google ADK, on Temporal for durable execution: The Agent Calls the Human. A fleet dispatch system escalates a disruption to an approver. We will intentionally kill the worker process mid-wait. Hours later, the human responds. State survives, and the agent resumes. The Human Calls the Agent. An operator interrupts a long-running task mid-flight to redirect it. The agent halts gracefully, surfaces state, accepts the override, and continues. You will leave with two production-ready architectural designs you can apply this week: agent-initiated approval gates with timeout and escalation semantics, and human-initiated interrupts with graceful agent halt and resumption.",
          "day": "Day 4 — Session Day 3",
          "time": "2:25pm-2:45pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.024037704,
        0.05377087,
        0.041573215,
        -0.033164173,
        0.023288475,
        -0.05232978,
        -0.05762111,
        -0.030180188,
        0.06745452,
        -0.2124457,
        0.010190336,
        -0.04708003,
        -0.013734995,
        -0.043787986,
        -0.06977005,
        0.037424803,
        0.09130636,
        -0.056857396,
        -0.08252962,
        -0.13218443,
        0.0008845177,
        -0.038674023,
        0.019692585,
        0.069657564,
        -0.062004246,
        0.137642,
        -0.055447955,
        0.055615313,
        -0.119462416,
        0.42606863,
        -0.078112766,
        0.030967833,
        0.04628538,
        -0.08994365,
        0.05895669,
        -0.10257924,
        -0.030891435,
        0.022371547,
        0.012699284,
        0.04222234,
        -0.048965577,
        0.0017147871,
        -0.027713079,
        0.081828505,
        -0.042533077,
        0.041036043,
        -0.09440681,
        -0.012059598,
        -0.046011902,
        -0.047103558,
        -0.058511667,
        0.004180795,
        0.053280264,
        -0.016588597,
        0.033218145,
        0.003638101,
        -0.047601074,
        -0.096749984,
        0.08866553,
        -0.031059507,
        -0.07448792,
        0.0141528165,
        -0.03536243,
        0.066199675,
        -0.040189218,
        -0.050040506,
        0.08651244,
        -0.0010493811,
        0.020168127,
        -0.037756052,
        0.009658665,
        0.0516649,
        -0.052989807,
        0.039749607,
        -0.03271165,
        -0.14974304,
        -0.02015976,
        0.07647241,
        -0.03276373,
        -0.01875342,
        -0.04950525,
        0.12904267,
        -0.07013006,
        -0.019364119,
        -0.015151881,
        -0.010253924,
        -0.064358644,
        0.061238654,
        -0.05004505,
        0.035177607,
        0.042539142,
        0.011007804,
        -0.08295591,
        -0.03245196,
        -0.03772395,
        0.052128702,
        -0.025213644,
        -0.00426876,
        -0.026746446,
        0.037611764,
        -0.05137132,
        -0.057017557,
        -0.0042472994,
        -0.0035007247,
        0.018942637,
        -0.046674155,
        -0.012505437,
        -0.011003162,
        0.042173274,
        0.004902205,
        -0.043441094,
        -0.62457025,
        0.0005516149,
        0.025041025,
        -0.0119275935,
        -0.16555868,
        0.0042249733,
        0.097508475,
        0.040967274,
        0.055230554,
        0.0057794373,
        0.048144657,
        -0.04109551,
        -0.02696531,
        0.009282008,
        0.005663527,
        0.04387672,
        0.028978983
      ]
    },
    {
      "id": "worldsfair-speaker-348",
      "conference": "worldsfair",
      "name": "Kay Malcolm",
      "role": "",
      "company": "Oracle",
      "twitter": "",
      "talks": [
        {
          "title": "No Memory, No Harness: Why the Database Is the Last Line of Defense",
          "description": "AI has made implementation faster, cheaper, and more widely available. That changes the real bottleneck in software.\n\nWhen every team can generate code, spin up agents, prototype workflows, and ship demos faster than ever, the advantage moves to a different layer: knowing what is worth building, who it is for, how it fits into the system, and what keeps it reliable in production.\n\nThe model is the easy part. Everything that makes an agent survive contact with production lives in the harness around it: orchestration, tooling, governance, and the memory core that keeps the system grounded when the model itself is probabilistic, forgetful, and non-deterministic. This talk walks the surface areas of an agent harness and consolidates the lessons we're learning as we ship them, from agentic applications in their current form (autonomous systems that now build their own automations) to the continual-learning loops that let agents improve from their own experience.\n\nWe'll look at how the discipline is segmenting. AI application development is no longer one role but several: agent engineers, memory engineers, and platform engineers. We'll map Oracle's primitives onto each as the current state of harness engineering takes shape. We'll also examine the two populations betting on this stack at once, enterprise customers who need governance, reliability, and scale, alongside the cracked developers who need fast, composable primitives, and why a well-engineered harness serves both. And we'll make the case that has held through every shift in the stack: memory isn't a feature you bolt on, it's the foundation the rest of the harness stands on. The database remains the memory core, and when everything above it is probabilistic, it's the last line of defense.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.08125457,
        0.099559315,
        0.02385739,
        0.039004616,
        0.02034217,
        -0.014458546,
        -0.02478357,
        -0.004671586,
        0.06110666,
        -0.18460408,
        -0.028171394,
        -0.02462081,
        -0.06279205,
        0.0018331158,
        -0.055244397,
        -0.0018845443,
        0.14455296,
        -0.06705504,
        -0.08193435,
        -0.08967043,
        0.010414537,
        -0.043827597,
        0.02109515,
        0.04381679,
        -0.10710604,
        0.08410712,
        -0.0363743,
        0.060982846,
        -0.08427757,
        0.48823786,
        -0.09865739,
        0.0395848,
        0.051023543,
        -0.035743408,
        0.05366461,
        -0.043398857,
        0.013104118,
        -0.018941676,
        0.011887674,
        0.045289528,
        -0.012175468,
        -0.002276087,
        -0.08161405,
        0.050224736,
        -0.051968817,
        -0.027817465,
        -0.046311468,
        -0.039174788,
        -0.026967624,
        -0.0693981,
        -0.069937825,
        -0.040838443,
        0.02471637,
        0.009012646,
        0.039025746,
        0.048088018,
        -0.00392234,
        -0.030204048,
        0.083479024,
        -0.06691455,
        -0.103440814,
        0.011717586,
        -0.054448385,
        0.002713896,
        0.008056905,
        -0.111790836,
        0.085639924,
        -0.054546617,
        -0.030001419,
        -0.042806953,
        0.015963893,
        -0.01259211,
        -0.07754956,
        -0.0073973397,
        -0.058937054,
        -0.09793224,
        -0.04050053,
        0.05830074,
        -0.054065444,
        0.016347192,
        -0.015074614,
        0.063885376,
        -0.09072261,
        -0.03405224,
        -0.0062040915,
        -0.027204826,
        -0.1288648,
        0.06390386,
        -0.02701178,
        0.01860709,
        0.055568,
        -0.0003070932,
        -0.011323963,
        0.017991044,
        -0.0965628,
        0.022613063,
        -0.064676836,
        0.0033852665,
        -0.027591012,
        0.06042191,
        -0.096489176,
        -0.07303273,
        0.035188958,
        -0.037272297,
        -0.03214184,
        -0.086010106,
        -0.052298848,
        -0.027300088,
        0.086457156,
        -0.0024885992,
        -0.007986242,
        -0.58770454,
        -0.005431613,
        -0.015453244,
        0.024178758,
        -0.12500085,
        0.02416139,
        -0.024521364,
        0.035405643,
        0.086108156,
        -0.021047466,
        0.09441095,
        -0.058038156,
        -0.014420131,
        0.013480558,
        0.02148577,
        0.063225165,
        -0.026628966
      ]
    },
    {
      "id": "worldsfair-speaker-349",
      "conference": "worldsfair",
      "name": "Ahmed Ahres",
      "role": "",
      "company": "Reactor",
      "twitter": "https://x.com/Boudatw",
      "talks": [
        {
          "title": "The Next Medium: Why Real-Time Interactive Video Changes Everything for Developers",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.04065121,
        -0.014174821,
        0.07047835,
        -0.0229134,
        0.0022125975,
        -0.06402233,
        -0.054558083,
        0.043926515,
        0.04475805,
        -0.23366109,
        0.027166974,
        -0.07074289,
        -0.029248478,
        -0.08291108,
        -0.080014214,
        0.015471154,
        0.07860434,
        -0.03153914,
        -0.07325753,
        -0.1291919,
        -0.042855002,
        -0.11378447,
        -0.032109033,
        -0.026798103,
        -0.11830747,
        0.1314977,
        -0.07775903,
        -0.02480238,
        -0.14537935,
        0.47241566,
        -0.05474642,
        -0.01876935,
        0.014582553,
        -0.10956356,
        0.047132604,
        -0.0060113748,
        -0.020629598,
        0.032844286,
        0.0137149915,
        0.0884024,
        -0.0062837983,
        0.055800665,
        0.016570637,
        0.032613944,
        0.028221523,
        0.040180825,
        -0.034348942,
        -0.041229807,
        -0.035250466,
        -0.05678907,
        -0.023121728,
        0.06923846,
        0.010395414,
        -0.030326365,
        0.021530801,
        -0.00073939876,
        0.009152105,
        -0.0640506,
        0.06927371,
        -0.075948425,
        -0.045692425,
        -0.04448645,
        0.070740946,
        0.027703315,
        0.015711294,
        -0.012475458,
        0.09394042,
        -0.047947858,
        0.020410363,
        -0.0334833,
        -0.031338617,
        -0.019716058,
        -0.023636349,
        -0.01608087,
        -0.028033232,
        0.007124763,
        0.08146784,
        0.08861248,
        0.009873902,
        0.0046009845,
        0.015320495,
        -0.016186459,
        -0.017649142,
        -0.05821849,
        0.02644488,
        -0.012898384,
        -0.0931199,
        0.045377966,
        0.039375685,
        0.099879496,
        0.034643456,
        0.06118566,
        0.0013456226,
        0.07472482,
        -0.026400944,
        -0.0052250335,
        0.010193519,
        0.047557555,
        0.12514529,
        0.03511677,
        0.047307394,
        0.0057377024,
        0.0012547373,
        -0.030849366,
        -0.005758774,
        -0.11685602,
        -0.012558754,
        -0.033038184,
        0.1260119,
        -0.036386248,
        -0.08918586,
        -0.5687445,
        0.012358113,
        -0.06045221,
        0.0069120172,
        -0.1597817,
        -0.040308613,
        -0.026622212,
        -0.015648903,
        0.068497725,
        -0.023995548,
        0.048785023,
        -0.014076006,
        0.05109098,
        0.02177935,
        0.08515669,
        -0.021110563,
        0.00060210295
      ]
    },
    {
      "id": "worldsfair-speaker-350",
      "conference": "worldsfair",
      "name": "Jay Mok",
      "role": "",
      "company": "PayPal",
      "twitter": "",
      "talks": [
        {
          "title": "Your Agent Just Authorized What?!",
          "description": "The nightmare scenario writes itself: your agent just ran off with your credit card and maxed it out on concert tickets, crypto, and a questionable NFT collection. Relax — we're building the guardrails. When an agent acts on your behalf, three questions must always be answerable: Did the human authorize this? Did they authorize this, now, in this scope? And can we prove it later? This talk maps three permissioning layers onto a stakes ladder: OAuth scopes at the bottom (broad capability, weak per-action proof, fine when reversible), Claude Code's tool-scoped allow/ask/deny model in the middle (brilliant for developer tooling, but no cryptographic evidence), and signed payment mandates at the top — where FIDO's Agentic Payments Working Group is building toward cryptographically-bound, constraint-carrying credentials. We'll share artifacts from Agent to Agent payments using our Shared Vault and Oauth to our constraint carrying Approval token leveraging our pillars of Identity and Buyer and Seller protection. You leave with a stakes × evidence matrix and a mental model that applies beyond payments: medical orders, e-signatures, securities trading, activities where you want you want to be more careful with your agent.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.017961288,
        0.041751746,
        -0.011065125,
        0.0030175804,
        0.016561715,
        0.01264743,
        -0.04554178,
        0.011486898,
        0.05261418,
        -0.22853836,
        -0.040147375,
        -0.07774648,
        -0.021721838,
        0.02293992,
        -0.046893526,
        -0.029135274,
        0.058239352,
        -0.09015181,
        -0.054487824,
        -0.014524101,
        -0.053679485,
        -0.098092414,
        0.04782668,
        -0.011432892,
        -0.051154457,
        0.086580396,
        0.041643605,
        0.12158554,
        -0.04301284,
        0.48235542,
        -0.08453892,
        0.063466474,
        -0.0039509274,
        -0.03869435,
        0.033076108,
        -0.06063831,
        -0.019535128,
        -0.0018185588,
        0.013989605,
        -0.00995462,
        0.025668003,
        -0.025596775,
        -0.056850847,
        0.08386073,
        -0.04111278,
        -0.03147492,
        -0.04974067,
        -0.013258236,
        0.0054371953,
        -0.031373397,
        -0.08848525,
        -0.026021738,
        0.014410739,
        -0.035220295,
        0.046102256,
        -0.028136253,
        -0.0038235507,
        -0.009194994,
        0.1384538,
        -0.080266275,
        -0.073598206,
        -0.0050090523,
        -0.022602249,
        0.011421303,
        -0.015741905,
        -0.083937585,
        0.035680197,
        -0.090308,
        0.073484756,
        -0.06103462,
        -0.0010488458,
        0.019385656,
        -0.06169748,
        0.052139863,
        -0.1143705,
        -0.10619478,
        -0.053825036,
        0.05243185,
        0.007776033,
        0.03417042,
        0.014793408,
        0.06831016,
        -0.079524346,
        -0.07797463,
        0.00613749,
        -0.027668696,
        -0.02681143,
        0.0022852127,
        0.0023476228,
        0.0742225,
        -0.03686884,
        -0.025742045,
        -0.054604,
        0.048920445,
        -0.079397656,
        0.04447591,
        -0.0734187,
        0.017118258,
        -0.018135553,
        0.006953077,
        -0.02347127,
        -0.12744638,
        0.0568883,
        0.00024015376,
        0.011255199,
        -0.019898022,
        0.020540142,
        -0.027515713,
        0.05508401,
        0.013560267,
        -0.037531085,
        -0.60719186,
        -0.01640949,
        -0.00191748,
        0.024357023,
        -0.14876626,
        0.072758056,
        0.016235964,
        0.00025703872,
        0.09990293,
        -0.055544786,
        0.0681911,
        0.005026054,
        -0.012926839,
        -0.045898695,
        -0.0019086682,
        0.05257596,
        -0.023154894
      ]
    },
    {
      "id": "worldsfair-speaker-351",
      "conference": "worldsfair",
      "name": "Idan Gazit",
      "role": "Senior Director of Research, GitHub Next",
      "company": "GitHub",
      "twitter": "",
      "talks": [
        {
          "title": "Realtime multiplayer, automation, and you!",
          "description": "Now that the models are powerful and the agents are capable, why are we still approaching software development as if it's the same activity that it used to be, but \"faster\"? GitHub Next thinks about what this future wants to be through two lenses: - Automation: intelligence allows us to automate much more than we could with heuristics alone. How should that automation work? What guardrails do we have to put in place so that our CISOs allow us to do that? - Collaboration: agents can understand anything in your codebase, but what about all the facts that are in the heads of your teammates? Whether it's corporate politics or taste, how do we get the humans to leak that context where agents can see it and use it to produce better outcomes? Realtime multiplayer tools have displaced every turn-based tool out there. What should that look like for code? It's not going to be as simple as multiple cursors. Come by to hear more about what GitHub Next is learning about the changing shape of software creation — one that allows us to build better, not merely faster. One that allows us to scale up teams, not only individuals. And one where automations buy us time for craft and polish, not slop. We were promised flying cars, instead we have fifteen terminals. Let's have a nicer future than that.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.004064789,
        0.04503746,
        0.016087262,
        -0.048948705,
        0.046312083,
        -0.033971917,
        -0.09368453,
        -0.014909165,
        0.021606697,
        -0.2175103,
        -0.035483755,
        -0.047146473,
        -0.024564596,
        -0.066871,
        -0.037676573,
        -0.026232462,
        0.08039126,
        -0.036101915,
        -0.008889641,
        -0.123647735,
        -0.019187203,
        -0.075061396,
        0.01655907,
        0.05929306,
        -0.050789833,
        0.1625328,
        0.04073334,
        0.04566375,
        -0.106233425,
        0.4280643,
        -0.08368248,
        -0.02142425,
        0.037755653,
        -0.07373654,
        0.08891887,
        -0.08025178,
        -0.041296933,
        -0.082270324,
        0.025106786,
        0.047508433,
        0.0005809262,
        0.029967569,
        -0.08916455,
        0.026257263,
        0.0061950698,
        -0.090987615,
        -0.12487086,
        0.034546137,
        -0.039803308,
        -0.053125445,
        -0.06638939,
        -0.041493967,
        0.052574754,
        -0.03774339,
        0.015866712,
        0.007250039,
        0.03573858,
        -0.12634134,
        0.06340557,
        0.02258347,
        -0.09441162,
        0.025017638,
        0.0074456595,
        0.018844265,
        0.043553464,
        -0.049814705,
        -0.0039797937,
        -0.011477769,
        -0.024863986,
        -0.02103473,
        0.013312114,
        0.10060722,
        -0.032680765,
        -0.022437878,
        -0.032897666,
        -0.07781954,
        -0.0033910053,
        0.096691795,
        -0.038654294,
        -0.04974282,
        0.008876877,
        0.008847181,
        -0.128835,
        0.024102433,
        0.031042963,
        0.011879623,
        -0.07408985,
        0.0020235954,
        0.015946463,
        0.039714336,
        0.036491375,
        -0.04564234,
        -0.05514106,
        0.048928574,
        -0.052203324,
        0.06868617,
        -0.052399628,
        0.030239917,
        0.034440372,
        -0.019301327,
        0.0011774193,
        -0.07308536,
        0.011714629,
        -0.03490715,
        -0.0247197,
        -0.062400635,
        0.008674346,
        0.016353806,
        0.060041677,
        -0.052272964,
        0.010041351,
        -0.5989449,
        0.029051078,
        -0.011135126,
        -0.035491318,
        -0.17925206,
        0.014971539,
        -0.013723004,
        0.08746832,
        0.10220544,
        -0.035703566,
        0.13429382,
        -0.029887259,
        0.008693924,
        -0.019143408,
        0.06557161,
        0.05642138,
        0.0053701205
      ]
    },
    {
      "id": "worldsfair-speaker-352",
      "conference": "worldsfair",
      "name": "Arjun Harbhajanka",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Quill: Event-Driven AI That Keeps Documentation Alive at Enterprise Scale",
          "description": "Documentation is where good intentions go to die. Teams write it once, then it decays silently as code evolves. What if documentation could maintain itself? Quill is an event-driven platform that uses LLMs to automatically generate, update, validate, and summarize software documentation, triggered by the same git events teams already produce. When a PR merges, Quill reads the code changes, identifies affected docs, and opens PRs with precise updates. When a release ships, it synthesizes commit history and deployment data into stakeholder-ready release notes. When docs violate quality standards, it auto-fixes what it can and flags what it can't. This talk dives into the AI engineering decisions that made Quill work in production at scale. We cover how we designed mode-aware prompts using the Diataxis documentation framework to give the LLM a structured target instead of open-ended prose. We explore the no change narration principle, why AI-updated docs must never say previously or was changed to, and how a single prompting constraint transformed output quality. We walk through the multi-pass generation pipeline: per-file LLM calls, aggregation into a cohesive corpus, and information architecture restructuring, and how we handle repos with hundreds of source files without blowing up context windows or costs. Beyond prompting, we cover the system design: an event gateway that fans out git webhooks and package lifecycle events to specialized microservices, a Neo4j knowledge graph that models relationships between repositories, packages, and deployments, and a per-repository configuration system that lets teams opt into exactly the AI actions they trust. Attendees will leave with concrete patterns for building reliable, controllable AI-powered developer tools, not demos, but production systems.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.0208381,
        0.05667808,
        0.0153739955,
        0.06567801,
        -0.00051033305,
        -0.011232183,
        -0.11382221,
        0.039289527,
        0.056741916,
        -0.2072668,
        -0.025224686,
        0.013013206,
        0.0010912142,
        -0.015195013,
        -0.00012775228,
        -0.064667195,
        0.046329886,
        -0.048206337,
        -0.083313696,
        -0.082279414,
        -0.062387913,
        -0.0075707766,
        -0.003319839,
        0.026196336,
        -0.05043153,
        0.1488195,
        0.03436732,
        0.07504436,
        -0.1224335,
        0.45302215,
        -0.068643644,
        -0.011868503,
        -0.0043771905,
        -0.039902534,
        0.058205746,
        -0.073832214,
        -0.02336984,
        -0.03505835,
        0.023348304,
        0.081154644,
        -0.024113026,
        0.050372113,
        -0.05708561,
        0.08749304,
        -0.012686312,
        -0.080397576,
        -0.098746195,
        0.02738986,
        0.03866745,
        -0.04745572,
        -0.0511629,
        0.0007611202,
        0.0042111934,
        -0.012336268,
        0.03837422,
        -0.016035736,
        0.010708764,
        -0.03359976,
        0.06993834,
        -0.037783984,
        -0.0599008,
        -0.020985091,
        0.02263015,
        0.003301798,
        -0.022810157,
        -0.026885753,
        0.107807934,
        -0.03853029,
        -0.014169277,
        -0.03050757,
        -0.024122087,
        -0.003965078,
        -0.040673107,
        0.03237523,
        -0.044295177,
        -0.114353836,
        -0.0829272,
        0.13633198,
        -0.06053008,
        0.010803575,
        -0.06974063,
        0.028802568,
        -0.048367526,
        -0.031891014,
        0.081700854,
        -0.0030998895,
        -0.047776617,
        0.018524827,
        0.0152051095,
        0.12522195,
        0.0037950552,
        -0.009453746,
        -0.0079819355,
        0.020263987,
        0.013415306,
        0.068106055,
        -0.02977639,
        0.06081631,
        -0.039339926,
        0.025960581,
        0.021430127,
        -0.05959283,
        0.031817507,
        0.021071358,
        0.030767212,
        -0.062414285,
        0.010350765,
        -0.028847845,
        -0.022159204,
        0.008493359,
        0.0012310263,
        -0.63514376,
        0.035722226,
        -0.044525113,
        -0.030202912,
        -0.14982258,
        0.055273034,
        -0.0022626866,
        0.02730548,
        0.06181537,
        -0.017836927,
        0.09640348,
        -0.04169737,
        -0.047425896,
        -0.0047867647,
        0.042677116,
        0.025908705,
        0.018993188
      ]
    },
    {
      "id": "worldsfair-speaker-353",
      "conference": "worldsfair",
      "name": "Justin Joyce",
      "role": "",
      "company": "Cloudflare",
      "twitter": "",
      "talks": [
        {
          "title": "Cloudflare AI in GTM with Justin Joyce",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0035917384,
        0.050905906,
        -0.005174388,
        -0.0053004166,
        0.070807025,
        -0.05546719,
        -0.10683647,
        0.04580122,
        0.0002430609,
        -0.2027115,
        -0.05654745,
        -0.057455752,
        0.046231125,
        -0.014609534,
        -0.022174282,
        0.0721168,
        0.12976892,
        -0.013465949,
        -0.0692582,
        -0.10789014,
        -0.02552789,
        -0.10858811,
        0.008546361,
        0.0030861055,
        -0.02687032,
        0.05857307,
        0.03490615,
        0.06963275,
        -0.11744348,
        0.43655148,
        -0.111762024,
        -0.012869333,
        0.027994027,
        0.0072968532,
        0.011983872,
        -0.062391754,
        -0.03463,
        0.023927206,
        0.056141824,
        0.07644596,
        -0.014288944,
        -0.022588957,
        -0.057996575,
        0.026614603,
        0.036725216,
        0.043828692,
        -0.073665075,
        -0.010150765,
        0.01749918,
        -0.026501376,
        -0.08857467,
        0.031429756,
        -0.011770878,
        -0.037893638,
        0.02249363,
        0.045881025,
        -0.020460183,
        -0.06948205,
        0.10136022,
        -0.09932894,
        -0.12378019,
        -0.06621805,
        -0.06511294,
        -0.020190833,
        0.02628545,
        -0.022712974,
        0.05355905,
        -0.013534331,
        -0.04728212,
        -0.07037221,
        -0.090887755,
        0.054819897,
        -0.050796345,
        0.0065794275,
        -0.07384233,
        -0.05450393,
        -0.014553861,
        0.05141421,
        0.020230755,
        -0.082546905,
        0.040125314,
        0.08683024,
        -0.07939318,
        -0.026258342,
        -0.04609858,
        -0.074112,
        -0.028400233,
        0.02487797,
        -0.054740548,
        0.052571725,
        0.03214528,
        -0.088235974,
        0.028848303,
        -0.008533789,
        -0.07620696,
        -0.0043420563,
        -0.047866713,
        0.043332044,
        -0.00892102,
        0.003184798,
        -0.061678365,
        -0.116243936,
        0.023740869,
        0.049395632,
        -0.038736984,
        -0.016706936,
        0.07606899,
        -0.040783003,
        0.026956717,
        0.024801863,
        -0.010678973,
        -0.61243886,
        0.053647723,
        0.012296462,
        0.034665354,
        -0.15777215,
        0.011102862,
        0.021777954,
        0.023803527,
        0.086982965,
        0.013950361,
        0.0023336962,
        -0.026319891,
        0.038843144,
        0.01065622,
        0.07804511,
        0.082096845,
        -0.0012606584
      ]
    },
    {
      "id": "worldsfair-speaker-354",
      "conference": "worldsfair",
      "name": "Dan Feng",
      "role": "",
      "company": "Maven Clinic",
      "twitter": "",
      "talks": [
        {
          "title": "How to build an AI-Native Health Company",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.07025506,
        -0.008821142,
        -0.056160513,
        0.0050139544,
        -0.0016611874,
        -0.011018074,
        -0.03701266,
        0.0071688853,
        0.039282136,
        -0.20936228,
        -0.00721592,
        -0.049449373,
        -0.009665273,
        0.038694058,
        -0.03479166,
        0.070288226,
        0.12771936,
        -0.059036758,
        -0.027750649,
        -0.08884107,
        -0.04274306,
        -0.081177026,
        -0.011900287,
        0.036568046,
        -0.045933057,
        0.07183527,
        0.025200125,
        0.07707762,
        -0.099174455,
        0.4500895,
        -0.11425686,
        0.022401461,
        0.0080687655,
        -0.026054973,
        0.040540252,
        -0.06496443,
        -0.016585818,
        -0.014026181,
        0.052314024,
        0.10031279,
        0.04221309,
        0.02433297,
        -0.012582623,
        0.042524397,
        -0.06478119,
        -0.020233555,
        -0.027705599,
        -0.01491284,
        -0.009505755,
        -0.020632016,
        -0.09956335,
        -0.035666585,
        0.015025678,
        -0.05451544,
        -0.0067522274,
        -0.011903062,
        0.0039199395,
        -0.057996716,
        0.05036547,
        -0.041273635,
        -0.16277815,
        0.041417617,
        0.04640459,
        -0.04292538,
        0.03307863,
        -0.07526535,
        -0.016418323,
        -0.048981983,
        -0.0039650225,
        -0.03280502,
        0.023560252,
        -0.0038781527,
        -0.037483,
        0.03016062,
        -0.11019737,
        -0.07479082,
        0.02982602,
        0.11573067,
        -0.08102079,
        0.0170099,
        0.042287216,
        0.061663218,
        -0.07599809,
        -0.0533725,
        -0.014785817,
        -0.020025672,
        0.001296078,
        -0.0090687135,
        -0.0025302777,
        0.08830056,
        0.07401261,
        0.05454495,
        -0.014286698,
        -0.00044719834,
        -0.08707891,
        0.08441631,
        -0.005172161,
        0.021411002,
        -0.077220514,
        0.036446355,
        0.03639039,
        0.029798653,
        0.020659022,
        -0.00866174,
        -0.023415115,
        -0.0070101316,
        -0.053327344,
        -0.028347192,
        0.07764659,
        0.0323976,
        -0.040968094,
        -0.61640674,
        0.006462634,
        0.0010842832,
        -0.046564724,
        -0.148129,
        0.03200779,
        -0.017687242,
        0.025545632,
        0.14401247,
        -0.03248076,
        0.044595,
        0.10684015,
        -0.07011601,
        0.022343287,
        0.026921613,
        0.040420957,
        -0.028490052
      ]
    },
    {
      "id": "worldsfair-speaker-355",
      "conference": "worldsfair",
      "name": "Asaf Gardin",
      "role": "Inference Engineer",
      "company": "AI21",
      "twitter": "",
      "talks": [
        {
          "title": "Two Bugs That Hid in Plain Sight: A vLLM Debugging Detective Story",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.022016304,
        0.0044687097,
        0.007043717,
        -0.010908157,
        0.03162938,
        -0.017726947,
        -0.03219199,
        0.008611244,
        0.024045313,
        -0.21288872,
        -0.019228972,
        -0.031144295,
        0.0031034886,
        -0.022915455,
        -0.022717068,
        0.017279347,
        0.14833428,
        -0.074058235,
        -0.039862685,
        -0.14476842,
        -0.023197902,
        -0.08070418,
        -0.06820432,
        0.023291824,
        -0.06488305,
        0.050418355,
        -0.004366976,
        0.04532823,
        -0.11675344,
        0.46429318,
        -0.11149677,
        -0.012461725,
        0.034202274,
        -0.06935369,
        0.07449046,
        -0.11919962,
        -0.00622236,
        -0.021580582,
        0.014399196,
        0.07544196,
        -0.0133938445,
        -0.01797126,
        -0.044826563,
        0.0154793635,
        -0.050056484,
        -0.05033624,
        -0.10910781,
        0.035499733,
        0.006709319,
        -0.09603823,
        -0.08537044,
        0.015050759,
        0.09988157,
        0.003677837,
        -0.030323574,
        0.0032183928,
        -0.03064767,
        -0.13150837,
        0.07937881,
        -0.037425507,
        -0.07470121,
        -0.006822635,
        -0.020036576,
        0.021459265,
        0.051096547,
        -0.0038783634,
        0.038227767,
        -0.08754609,
        -0.041581277,
        0.033531632,
        -0.015431884,
        -0.014147458,
        -0.03755881,
        0.012181131,
        -0.10134869,
        -0.06272278,
        -0.01615906,
        0.118829265,
        -0.00975398,
        -0.03457016,
        0.01967528,
        0.032336675,
        -0.07288763,
        0.018127052,
        0.050328843,
        -0.03158194,
        -0.034396425,
        0.07364005,
        -0.02873082,
        0.058371738,
        0.03732134,
        -0.0787667,
        -0.04150821,
        0.051463477,
        -0.060312066,
        0.03035279,
        -0.036439665,
        0.06396654,
        0.021930346,
        -0.0395082,
        0.0048809005,
        -0.046078034,
        0.018535752,
        -0.0041125575,
        -0.053070188,
        -0.07032144,
        -0.041488342,
        -0.00016559396,
        0.07442585,
        -0.015280936,
        -0.08610869,
        -0.58514863,
        0.00012407888,
        -0.046217736,
        -0.09858771,
        -0.15740484,
        0.020890106,
        -0.00629594,
        -0.0026383882,
        0.10717841,
        -0.03830346,
        0.040756058,
        -0.031092195,
        -0.007576789,
        0.036163643,
        0.04133494,
        0.0831202,
        -0.0054642064
      ]
    },
    {
      "id": "worldsfair-speaker-356",
      "conference": "worldsfair",
      "name": "Salman Munaf",
      "role": "Lead Site Reliability Engineer",
      "company": "TikTok",
      "twitter": "",
      "talks": [
        {
          "title": "AI Agents Are Just Distributed Systems Now",
          "description": "AI agents are often described as a new kind of software, but once they move beyond chat and start calling tools, reading data, making decisions, retrying tasks, and coordinating workflows, they begin to look a lot like distributed systems. They have state. They call external services. They depend on APIs. They fail partially. They retry. They time out. They can loop. They can act on stale context. They can produce inconsistent results. And when something goes wrong, teams need logs, traces, permissions, ownership, and rollback paths just like they do with any other production system. This session will give engineers a practical way to reason about AI agents using familiar distributed systems concepts. We will break down the agent loop: planning, tool use, observation, memory, and retries. Then we will map common agent failure modes to engineering patterns teams already know, including timeouts, circuit breakers, idempotency, rate limits, least privilege, observability, and human approval. The goal is to move past the hype and treat agents like real production systems. Attendees will leave with a clear mental model for designing, debugging, and operating agents safely, especially as they become part of customer-facing products, internal developer tools, and business workflows.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06272754,
        0.04148185,
        0.019945294,
        -0.007762275,
        0.029084524,
        -0.026357751,
        -0.0071475613,
        -0.040482506,
        0.061854936,
        -0.25546658,
        0.00014385559,
        -0.047195785,
        -0.0031144558,
        -0.087884344,
        -0.060000505,
        0.018191205,
        0.12816891,
        -0.07063478,
        -0.041746445,
        -0.10100289,
        0.0034082402,
        -0.07414683,
        0.015395181,
        0.045726866,
        -0.021894941,
        0.16645327,
        -0.003549114,
        0.035420366,
        -0.057903484,
        0.4687994,
        -0.10378009,
        0.016048739,
        0.04015051,
        -0.066088684,
        0.06299103,
        -0.031034973,
        -0.06284612,
        0.026293661,
        0.006531559,
        0.030761218,
        -0.042139456,
        0.050991356,
        -0.03394136,
        0.04072934,
        -0.038309794,
        -0.049009804,
        -0.053144198,
        0.057518702,
        0.013575389,
        -0.0037417675,
        -0.0073846737,
        -0.006998053,
        0.051416717,
        0.008169455,
        0.019342832,
        -0.00540239,
        -0.0007042241,
        -0.030474303,
        0.07964974,
        -0.012422533,
        -0.08313871,
        0.040631533,
        0.019323247,
        0.014289783,
        -0.0053533,
        -0.12401156,
        0.0807789,
        -0.0313838,
        -0.032542843,
        -0.051676154,
        0.0040515247,
        0.030154223,
        -0.033608414,
        -0.019364605,
        -0.074468285,
        -0.14520286,
        0.0040045963,
        0.06934381,
        -0.020334156,
        0.017353002,
        0.01583085,
        0.0838786,
        -0.11733621,
        -0.019886976,
        -0.01173804,
        -0.021814883,
        -0.083957136,
        0.05393183,
        -0.013487944,
        0.082467936,
        0.019157069,
        0.0015069138,
        -0.031776734,
        0.027646275,
        -0.10652219,
        0.06427581,
        -0.056342468,
        0.03815028,
        -0.046531875,
        0.06622242,
        0.0184787,
        -0.014671562,
        0.04747098,
        0.0034253323,
        -0.010181976,
        0.012880275,
        0.01754479,
        -0.025465678,
        0.10655786,
        -0.020544259,
        -0.04857085,
        -0.5904029,
        0.016283382,
        -0.007207887,
        -0.024405308,
        -0.14140682,
        0.042401936,
        0.024261093,
        0.07934917,
        0.1115384,
        -0.015051458,
        0.076762475,
        -0.06305364,
        0.0010418046,
        -0.015580476,
        0.024261087,
        0.015617548,
        0.011906828
      ]
    },
    {
      "id": "worldsfair-speaker-357",
      "conference": "worldsfair",
      "name": "Sebastian Fox",
      "role": "",
      "company": "Composo",
      "twitter": "",
      "talks": [
        {
          "title": "Inside 847 Production Clinical AI Notes",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.036899857,
        0.052019034,
        0.0024388868,
        0.035447847,
        0.01851232,
        0.03913295,
        -0.10695697,
        0.038696647,
        0.034771454,
        -0.19235536,
        -0.038793057,
        -0.027012363,
        0.008190207,
        -0.024376245,
        -0.050825685,
        0.036950096,
        0.09015264,
        -0.047661748,
        -0.027661294,
        -0.092449084,
        -0.0011676045,
        -0.051313095,
        -0.08970837,
        0.076739624,
        -0.11190208,
        0.049649887,
        -0.01747867,
        -0.020475937,
        -0.08475683,
        0.52072823,
        -0.10459339,
        0.041592196,
        0.0028601128,
        -0.0060286904,
        0.08702953,
        -0.053300116,
        0.0061438843,
        0.033937585,
        0.0394796,
        0.03427191,
        -0.033700105,
        0.0525201,
        -0.042918902,
        0.0014943328,
        0.081803456,
        -0.011228939,
        -0.063562594,
        -0.020736942,
        -0.047658473,
        -0.009663533,
        -0.06472481,
        0.015096333,
        -0.026403453,
        -0.04895478,
        -0.0039702887,
        -0.0029141284,
        0.012091249,
        0.0026983893,
        0.04475409,
        -0.026189316,
        -0.09508778,
        -0.059137236,
        0.029340735,
        0.011644643,
        0.04187129,
        -0.062884115,
        0.09933481,
        -0.08385314,
        0.01820071,
        -0.068723746,
        0.043397054,
        0.04358352,
        -0.060443215,
        -0.00083242974,
        -0.082787916,
        -0.036065456,
        -0.061223958,
        0.077731,
        0.013929494,
        0.017216923,
        0.0023399128,
        -0.04096068,
        -0.054954674,
        -0.05040537,
        0.045847625,
        0.022859829,
        -0.048793,
        0.017713819,
        0.0522281,
        0.052711613,
        0.06134681,
        0.00086989335,
        0.067141846,
        -0.035736687,
        0.0067136167,
        -0.053857442,
        -0.045574863,
        0.025861269,
        -0.027554676,
        0.001955406,
        -0.0071451864,
        -0.017973343,
        -0.008959953,
        -0.0055276365,
        -0.030947043,
        -0.027128175,
        -0.022426777,
        -0.08935849,
        0.10396663,
        0.030662982,
        -0.019934656,
        -0.61454505,
        -0.033092495,
        -0.054913983,
        -0.0068095746,
        -0.15554403,
        0.025877936,
        0.028063733,
        0.035728853,
        0.10698704,
        0.012294843,
        0.013823947,
        -0.015989276,
        0.015584985,
        0.036570445,
        0.026811812,
        0.02180119,
        -0.011167979
      ]
    },
    {
      "id": "worldsfair-speaker-358",
      "conference": "worldsfair",
      "name": "Dru Knox",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Harness Engineering: The New Core Skill for Agentic Developers",
          "description": "Harness engineering is emerging as a new core competency for agentic engineers. Your job isn't writing good code, it's upgrading your codebase so that agents reliably succeed. This talk covers the core loop of harness engineering, the most common codebase modifications you'll make, and how to 10x your harness engineering efforts with Tessl's harness engineering agent.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.06135461,
        0.09077327,
        0.12046181,
        0.048583277,
        0.046346676,
        -0.04790864,
        -0.06113221,
        -0.018236121,
        0.029196074,
        -0.20833693,
        -0.010989567,
        0.029412236,
        -0.0060128164,
        -0.005752599,
        -0.05532308,
        -0.033049278,
        0.17603889,
        -0.07545782,
        -0.0982922,
        -0.11259945,
        -0.01828574,
        -0.09842255,
        -0.019808052,
        0.056342106,
        -0.037937846,
        0.1147782,
        -0.0012405947,
        0.099390306,
        -0.09329299,
        0.42203087,
        -0.045151997,
        0.019860609,
        0.0073232306,
        -0.027782341,
        0.044826012,
        -0.035334744,
        -0.0282929,
        -0.01824809,
        0.046328828,
        0.048362188,
        0.0037130516,
        0.0461393,
        -0.056245327,
        0.035141792,
        -0.01802959,
        -0.029661492,
        -0.10477137,
        0.014323688,
        -0.009810597,
        -0.027456015,
        -0.07621231,
        0.014705864,
        0.045388877,
        -0.023196986,
        0.0586337,
        0.023287343,
        -0.025470808,
        -0.094722144,
        0.08141696,
        0.01827546,
        -0.04309248,
        0.055109754,
        -0.033911753,
        0.008373831,
        -0.029548585,
        -0.05101829,
        0.082152046,
        -0.06800919,
        0.0005329177,
        -0.10608261,
        0.020202808,
        0.060374577,
        -0.052740943,
        0.051366635,
        -0.032073073,
        -0.1476181,
        -0.019352071,
        0.08425097,
        -0.06058533,
        0.06587489,
        -0.055775173,
        0.057250597,
        -0.10401109,
        -0.07121578,
        -0.018634008,
        0.002864799,
        -0.021020679,
        0.024717681,
        -0.040651534,
        0.047695998,
        0.06593344,
        0.042467847,
        -0.031743355,
        0.011174772,
        -0.07774769,
        0.012310107,
        -0.055644576,
        0.04071987,
        -0.07491456,
        0.0028591538,
        0.023090284,
        -0.024290519,
        -0.01651953,
        0.016500624,
        0.02586606,
        -0.05271928,
        -0.014732704,
        0.03229911,
        0.033850722,
        0.03459108,
        0.016885689,
        -0.6171819,
        -0.017563626,
        0.013988855,
        -0.05784676,
        -0.1421641,
        0.0056190137,
        0.03235362,
        0.027438711,
        0.06816141,
        -0.008193239,
        0.07114235,
        -0.05752378,
        -0.019615298,
        0.0369705,
        0.04400205,
        0.06384195,
        0.0027981265
      ]
    },
    {
      "id": "worldsfair-speaker-359",
      "conference": "worldsfair",
      "name": "Felipe Blanes",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Designing Evals That Earn User Trust",
          "description": "Most teams measure their agent against a benchmark, ship it, and hope. But when your agent serves real users, a benchmark won't tell you if it's actually working. This session is about building an eval suite that captures what success looks like in production, runs against real user workflows, and feeds back into product decisions. Here's the flywheel we use in practice: start with what success looks like from the user's perspective, instrument production workflows to capture those signals, diagnose where the agent falls short, and feed those insights into the next thing you build. You'll see how it shaped concrete product bets, turning eval results from a report card into a discovery tool.",
          "day": "Day 4 — Session Day 3",
          "time": "2:50pm-3:10pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.040927093,
        0.053586192,
        -0.007125841,
        -0.011989032,
        -0.023454955,
        -0.06608367,
        -0.01580503,
        -0.02204242,
        0.018834613,
        -0.22905919,
        0.05602612,
        -0.052619085,
        -0.001188353,
        0.007221626,
        -0.06194157,
        0.04418995,
        0.12078804,
        -0.08037851,
        -0.064594865,
        -0.10689448,
        -0.052677207,
        -0.09789435,
        -0.024388717,
        0.06841525,
        0.010823913,
        0.061154466,
        0.033454042,
        0.085409135,
        -0.15268277,
        0.45323277,
        -0.1098304,
        0.016401377,
        0.020040728,
        -0.03730926,
        -0.015540933,
        -0.0029510546,
        -0.015138502,
        0.06169666,
        -0.035711348,
        0.01806103,
        0.009302907,
        0.12006212,
        -0.065740876,
        0.0626431,
        -0.03175875,
        -0.03722966,
        -0.057941128,
        0.022358932,
        -0.02326262,
        0.004588471,
        -0.053602796,
        -0.023306409,
        0.025666589,
        0.010915452,
        0.02598643,
        0.04928977,
        -0.015094537,
        -0.05599772,
        0.11372264,
        0.020506509,
        -0.05058489,
        -0.007124982,
        -0.013754348,
        -0.028402587,
        0.058429077,
        -0.07813607,
        0.09948117,
        -0.044371475,
        0.010659419,
        -0.07851759,
        0.04879201,
        -0.013070304,
        -0.013414623,
        -0.023124741,
        -0.053121287,
        -0.16150211,
        -0.062902555,
        0.07648469,
        -0.033232994,
        0.008229768,
        0.029773133,
        0.092328206,
        0.029811028,
        -0.032340292,
        0.033482056,
        0.05242426,
        -0.040945854,
        0.06979401,
        -0.06845607,
        0.04107138,
        0.013924535,
        0.0017211194,
        -0.08180397,
        -0.06125935,
        -0.028740218,
        0.035024107,
        -0.051544674,
        0.013719785,
        -0.016311709,
        0.038256723,
        -0.0081989085,
        0.013680366,
        -0.020655528,
        0.0070518847,
        0.026537662,
        -0.100547455,
        0.011121803,
        0.0611948,
        0.048514966,
        0.007763108,
        -0.029218368,
        -0.6040134,
        -0.0056643425,
        -0.0766412,
        -0.031504042,
        -0.15280053,
        0.041043337,
        0.026387075,
        -0.0030088732,
        0.04967234,
        0.0062562027,
        0.08562384,
        -0.022895664,
        0.041140128,
        -0.013213405,
        0.083284475,
        0.025900325,
        0.018635279
      ]
    },
    {
      "id": "worldsfair-speaker-360",
      "conference": "worldsfair",
      "name": "Andrew Qu",
      "role": "",
      "company": "Vercel",
      "twitter": "https://x.com/andrewqu",
      "talks": [
        {
          "title": "We Solved Agent Building - The Evolution of Building A Successful Data Science Agent",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.008274262,
        0.049684145,
        -0.011424962,
        0.0055987956,
        -0.012673057,
        0.012820954,
        -0.10442127,
        -0.009020499,
        -0.011718609,
        -0.24705885,
        -0.066072375,
        -0.059507344,
        0.07299423,
        -0.007967293,
        -0.09245706,
        0.034997772,
        0.053205945,
        -0.10639269,
        -0.0464048,
        -0.047787633,
        -0.07228756,
        -0.057979744,
        -0.0076335464,
        0.016057651,
        -0.03502193,
        0.13538145,
        -0.029112794,
        0.0626691,
        -0.13524888,
        0.46974733,
        -0.08128221,
        -0.032185256,
        -0.0038965708,
        0.0024969862,
        -0.014804227,
        -0.047180455,
        0.04583272,
        -0.016174462,
        0.017375283,
        0.06996241,
        0.022044025,
        0.0035711178,
        -0.061179616,
        0.060691062,
        -0.011407797,
        0.026077671,
        -0.054028917,
        -0.05913434,
        -0.053057585,
        -0.009365353,
        -0.037804224,
        0.02714981,
        0.074059516,
        -0.047904927,
        -0.0075775953,
        -0.01438519,
        -0.028258702,
        -0.0727025,
        0.04108472,
        -0.006955471,
        -0.12568206,
        -0.0510073,
        -0.0013373403,
        0.0018271691,
        -0.0095897345,
        -0.078872636,
        0.04424725,
        -0.058018923,
        -0.021737913,
        -0.08040948,
        -0.024451565,
        -0.009105623,
        -0.051367335,
        0.03652444,
        -0.10310814,
        -0.06914631,
        0.023531597,
        0.0678359,
        -0.09330888,
        -0.027526643,
        -0.01227379,
        0.031888634,
        -0.07578557,
        -0.07612553,
        0.03133084,
        -0.03544564,
        -0.056657873,
        0.049336486,
        -0.055533107,
        0.10199394,
        0.036856722,
        -0.012997823,
        -0.005504638,
        -0.04332223,
        -0.110511735,
        0.023583788,
        -0.082630455,
        0.01798811,
        0.04840906,
        -0.023379132,
        -0.016535241,
        0.00075000996,
        0.041446507,
        -0.019116746,
        -0.058464922,
        -0.079118475,
        0.014984027,
        -0.022355441,
        0.06866683,
        0.024310563,
        -0.06839577,
        -0.59482944,
        -0.055063028,
        -0.09043729,
        -0.0030334115,
        -0.14796314,
        -0.0073772958,
        -0.014405503,
        0.07703157,
        0.009454835,
        -0.038827762,
        0.06496971,
        -0.00015996383,
        0.01733936,
        -0.026918657,
        0.0019758334,
        0.046634454,
        -0.018328618
      ]
    },
    {
      "id": "worldsfair-speaker-361",
      "conference": "worldsfair",
      "name": "Brennan Erbz",
      "role": "CEO & Co-Founder",
      "company": "Flik",
      "twitter": "https://x.com/BrennanErbz",
      "talks": [
        {
          "title": "Beyond Prompts: Building a Multi-Agent Creative Computer That Orchestrates 5+ AI Models in Real-Time",
          "description": "Flik is a production multi-agent system that generates complete work, not pieces. This talk demonstrates how the system orchestrates Claude, Gemini, Nano, Seedance, and Eleven Labs in a single workspace across text, image, video, and audio; shows an end-to-end workflow from prompt to finished output; explains coordination across modalities; and covers built-in likeness/IP safety plus real customer examples.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.10478992,
        0.033191744,
        0.039029807,
        -0.041133527,
        -0.012010541,
        -0.053707447,
        -0.051041488,
        0.020847596,
        0.018234532,
        -0.20570308,
        -0.024201177,
        0.0023274638,
        0.02759148,
        0.025855169,
        -0.045172963,
        0.02538504,
        0.06940963,
        -0.06514747,
        -0.015091255,
        -0.09723972,
        -0.0026911572,
        -0.09518781,
        -0.010556267,
        0.017325891,
        -0.027600084,
        0.1132848,
        0.035767492,
        0.030700179,
        -0.06529068,
        0.4741457,
        -0.129283,
        0.007259686,
        0.03647886,
        0.004747815,
        -0.00324395,
        -0.048224285,
        -0.082388796,
        -0.03473114,
        -0.004329042,
        0.011046641,
        0.021577135,
        0.035157353,
        -0.03514821,
        0.019930504,
        0.03220767,
        -0.018374372,
        -0.0130397,
        -0.012847005,
        0.0072753215,
        0.016950125,
        0.019041613,
        0.010848271,
        0.07651652,
        -0.0049320073,
        0.05575353,
        0.011622479,
        0.07721369,
        -0.10432553,
        0.08630208,
        -0.028598066,
        -0.124476366,
        0.0007052998,
        0.03996687,
        0.05543022,
        -0.014810734,
        -0.122687265,
        0.03398031,
        -0.007020729,
        -0.021325333,
        -0.07011216,
        0.0040473975,
        0.05183696,
        -0.041550923,
        0.001987947,
        -0.028815115,
        -0.15508991,
        -0.017962145,
        0.05367229,
        0.02995444,
        -0.055457175,
        -0.017499816,
        0.05496187,
        -0.11406771,
        -0.10089253,
        -0.06437043,
        -0.04880777,
        -0.063983485,
        0.028687105,
        0.009182626,
        0.08839471,
        -0.015091987,
        0.04160649,
        0.0025100254,
        0.059633013,
        -0.07457432,
        0.016461655,
        -0.0080300355,
        0.013684313,
        -0.00033772184,
        0.03869911,
        0.020831965,
        -0.01945288,
        0.03910429,
        -0.0013200974,
        -0.0011207279,
        -0.026012098,
        0.08299982,
        0.034952577,
        0.050731257,
        -0.015647762,
        -0.027632473,
        -0.6109026,
        -0.021093763,
        -0.023545831,
        -0.048783306,
        -0.15272656,
        0.016252955,
        0.019056372,
        0.07589826,
        0.02528404,
        -0.097245984,
        0.09647776,
        -0.028674107,
        -0.04193831,
        0.019941706,
        0.04742956,
        0.079604276,
        -0.005450682
      ]
    },
    {
      "id": "worldsfair-speaker-362",
      "conference": "worldsfair",
      "name": "Gus Iwanaga",
      "role": "Product, UX, and Engineering lead for mosAIc",
      "company": "commercetools",
      "twitter": "https://x.com/guhgoi",
      "talks": [
        {
          "title": "The End of the Static Screen: Architecting Intent-Driven UX with Agentic Orchestration",
          "description": "For 30 years, interfaces were designed ahead: wireframes, fixed flows, pre-built dashboards - because we couldn't make them otherwise. Three shifts changed the constraint: LLMs that reason over business context, agentic frameworks that work at production grade, and composable backends that expose a real tool surface. With all three in place, the interface stops being something you design and ships as the output of an orchestrator composing it per intent. I'll walk through the hypothesis, the architecture we're running in production for enterprise commerce, and a live demo where it all moves.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.024463369,
        0.09391171,
        -0.01349007,
        -0.020933483,
        0.032509252,
        -0.058715124,
        -0.041452996,
        -0.062387407,
        0.03937773,
        -0.2038683,
        -0.008046929,
        -0.087100305,
        0.038213614,
        -0.039489936,
        -0.03356902,
        0.02941645,
        0.041678805,
        -0.039973665,
        -0.035409044,
        -0.097285084,
        -0.037527252,
        -0.05848476,
        -0.01632396,
        0.05947864,
        -0.027598305,
        0.056645088,
        0.02724376,
        0.030853633,
        -0.13278654,
        0.47260943,
        -0.058783874,
        0.063541286,
        -0.0054894206,
        -0.045262787,
        0.031235183,
        -0.019562423,
        -0.055173222,
        -0.0020865423,
        0.046213847,
        0.033597443,
        -0.036597304,
        0.04852648,
        -0.10555392,
        -0.0013349387,
        -0.03590069,
        -0.061018907,
        -0.09157418,
        -0.05305482,
        0.031201746,
        0.018446725,
        -0.13926534,
        -0.013503505,
        0.06724884,
        -0.022659713,
        0.019075062,
        -0.04149741,
        0.012138411,
        -0.05410444,
        0.064901985,
        -0.00036458203,
        -0.08894588,
        0.017804658,
        0.026061038,
        0.015427991,
        0.02163309,
        -0.053200673,
        0.08487415,
        -0.040169045,
        -0.0005371628,
        -0.09869717,
        0.01488453,
        0.05393657,
        -0.11172264,
        0.01946218,
        -0.05993038,
        -0.113394886,
        -0.11938376,
        0.036969222,
        0.037229482,
        -0.017232478,
        -0.045303352,
        0.002696567,
        -0.06665892,
        -0.020100042,
        -0.008084812,
        -0.023887668,
        0.01985167,
        0.028160209,
        -0.0063890964,
        0.08348802,
        0.0010009374,
        -0.016506817,
        -0.036948748,
        0.0004365671,
        -0.07740435,
        0.037078403,
        -0.015385981,
        -0.027679149,
        -0.027305642,
        0.02817649,
        -0.07759661,
        -0.08898506,
        -0.023972772,
        -0.019668458,
        -0.03452818,
        -0.05995745,
        0.042221718,
        -0.015088541,
        0.06004634,
        0.027348822,
        -0.011041602,
        -0.6254936,
        -0.01101968,
        0.05913089,
        0.026513433,
        -0.1495627,
        0.029373417,
        -0.03244053,
        0.037185527,
        0.08072098,
        0.012516127,
        0.05849355,
        -0.026533084,
        -0.04212104,
        0.020282157,
        0.038179897,
        0.031006353,
        0.017137164
      ]
    },
    {
      "id": "worldsfair-speaker-363",
      "conference": "worldsfair",
      "name": "Yogendra Miraje",
      "role": "Lead Machine Learning Engineer",
      "company": "FactSet",
      "twitter": "https://x.com/YogiNotTheBear",
      "talks": [
        {
          "title": "It's a Skill Issue : Best practices in building skills that work",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.000037125537,
        0.073536135,
        0.028564045,
        0.048608504,
        0.03175677,
        -0.028978767,
        -0.064249866,
        0.019556468,
        -0.060784,
        -0.23078535,
        -0.044148892,
        -0.042185083,
        0.029401943,
        -0.00591801,
        -0.055281933,
        -0.020764742,
        0.089441195,
        -0.07316546,
        -0.023960793,
        -0.13204747,
        -0.08558454,
        -0.07323027,
        -0.003167597,
        0.02237148,
        -0.017683115,
        0.040150866,
        -0.04403496,
        0.04698034,
        -0.071346216,
        0.49657628,
        -0.05878459,
        -0.005328197,
        0.006908573,
        -0.0602234,
        0.08049334,
        0.004977501,
        -0.013485383,
        0.025227075,
        -0.022498539,
        0.08312595,
        -0.015397495,
        0.05778792,
        -0.04725035,
        0.036108673,
        0.00968596,
        0.022074101,
        -0.1101074,
        0.06151701,
        0.009516609,
        -0.07073155,
        -0.067110315,
        -0.012678852,
        0.0046235286,
        -0.027557757,
        0.009837173,
        0.012381061,
        0.0069482196,
        -0.089487605,
        0.103241704,
        -0.020803392,
        -0.048513927,
        -0.018161347,
        -0.05232472,
        -0.016675241,
        0.0160904,
        -0.01093155,
        0.08993308,
        -0.085627906,
        -0.013100461,
        0.07697562,
        0.023764256,
        -0.05099337,
        -0.014778792,
        0.077978656,
        -0.062051825,
        -0.03607354,
        -0.0071750227,
        0.07717873,
        -0.018621447,
        0.029015763,
        0.012768495,
        0.054534752,
        -0.10149005,
        -0.048637703,
        -0.052107487,
        -0.055589426,
        -0.06979272,
        0.028706428,
        -0.055023145,
        0.10264318,
        0.010539707,
        -0.022258358,
        -0.025140096,
        -0.016289033,
        -0.05220412,
        0.018496173,
        -0.11586372,
        0.0545852,
        -0.049431052,
        -0.020764353,
        -0.009329883,
        -0.0059464523,
        -0.009516367,
        -0.01032028,
        -0.008708925,
        -0.082432866,
        0.026838152,
        -0.013710753,
        0.04398622,
        -0.014330525,
        -0.06416175,
        -0.60514677,
        -0.018990308,
        -0.015953626,
        -0.03826946,
        -0.17697611,
        -0.002982226,
        -0.03467033,
        -0.012108304,
        0.067738466,
        -0.04598991,
        0.054153275,
        -0.019305702,
        -0.018244112,
        -0.035059102,
        0.024276193,
        0.025208436,
        -0.0010055108
      ]
    },
    {
      "id": "worldsfair-speaker-364",
      "conference": "worldsfair",
      "name": "Matt Dailey",
      "role": "",
      "company": "Ref.",
      "twitter": "",
      "talks": [
        {
          "title": "Velocity Sickness: What Happens When Your Whole Team Gets 10x Faster",
          "description": "Learn more about Ref: https://ref.tools/\n\nAI made writing code nearly free, and on most teams, that's quietly breaking how the team works. Individually, everyone feels ten times faster. Together, the signals point the other way: too many PRs moving in too many directions, engineers throwing away whole agent sessions and starting over (\"declaring agent bankruptcy\"), and critical decisions getting made inside agent chats that no one will ever see or review. There's a lot of energy, and it's all going somewhere different. I call this velocity sickness: the organizational pain that comes from individual speed. It's the engineering version of an author who ships a book a week: prolific, productive, and completely unreadable by the team that's supposed to build on it.\n\nAlmost every conversation about AI coding is about making one engineer faster. This talk is about what happens to the team when all of them are. Once implementation stops being the bottleneck, the hard part isn't writing the code. It's tracking it, reviewing it, and keeping a hundred parallel decisions coherent. That's the problem eng leaders are actually being handed, and it's the one this session takes on directly.\n\nEngineering has always had three phases: plan, implement, polish. AI collapsed the middle one to almost nothing, so the leverage, and the real work, move to the decision-heavy ends. The fix isn't better prompts; it's changing what our tools treat as first-class. We have to split the decision layer from the implementation layer: humans spend their time at the decision layer, reviewing and making the choices that matter, while agents handle the implementation. That means durable, reviewable plans, not ephemeral chats. Review the decisions before you review the diff.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.011081947,
        -0.011968141,
        0.06954233,
        0.024004498,
        -0.039146826,
        0.059109706,
        -0.077501416,
        0.05002835,
        0.07593988,
        -0.19760096,
        -0.04558524,
        -0.06809938,
        0.012135473,
        0.0041987444,
        -0.021343874,
        -0.016181452,
        0.08130279,
        -0.08316363,
        -0.06158277,
        -0.058195736,
        0.01910149,
        -0.07925925,
        0.011218447,
        0.01701917,
        0.005553611,
        0.1154963,
        -0.008567869,
        0.081048645,
        -0.10365903,
        0.4827045,
        -0.060976546,
        0.040873874,
        0.015605736,
        0.017844426,
        0.0621206,
        -0.09345561,
        -0.061998878,
        -0.03186037,
        -0.04287842,
        0.080880195,
        0.009257614,
        0.027127942,
        -0.11625282,
        -0.0019887616,
        -0.008594546,
        -0.030954272,
        -0.04003534,
        -0.04176843,
        -0.041156534,
        -0.078758046,
        -0.06514995,
        0.051555354,
        -0.05235732,
        -0.06152823,
        0.007020656,
        0.05742867,
        -0.033166993,
        -0.10044851,
        0.016953086,
        -0.013240548,
        -0.09462146,
        0.051914927,
        -0.002954355,
        0.009443274,
        -0.014122227,
        -0.09293893,
        0.002526654,
        -0.018791785,
        -0.042465393,
        -0.063011535,
        0.0002500368,
        0.05490619,
        -0.08262218,
        -0.0050318637,
        -0.026381528,
        -0.08262702,
        -0.08357222,
        0.07780393,
        -0.08373291,
        0.04719974,
        -0.060464427,
        0.011125585,
        -0.14287947,
        0.057632167,
        0.014977244,
        -0.010549671,
        -0.013013517,
        0.034587584,
        -0.0142614655,
        0.06140244,
        0.042421,
        -0.046445142,
        -0.05736948,
        0.024113288,
        -0.099698104,
        0.061230537,
        0.048029438,
        -0.004058281,
        -0.03671718,
        -0.011994412,
        -0.004120438,
        -0.07765342,
        0.018841779,
        -0.038568143,
        -0.03675378,
        -0.07155938,
        -0.014094667,
        -0.03169799,
        0.03160553,
        -0.041624527,
        -0.009818891,
        -0.594204,
        0.014979652,
        -0.005663445,
        0.013140105,
        -0.14717822,
        0.019980146,
        0.05880406,
        0.054016184,
        0.08783113,
        -0.0011776063,
        0.06688662,
        -0.009683558,
        -0.012149997,
        0.03580998,
        0.068286,
        0.07998082,
        0.00035880037
      ]
    },
    {
      "id": "worldsfair-speaker-365",
      "conference": "worldsfair",
      "name": "Bil Erdenekhuyag",
      "role": "",
      "company": "Intercom",
      "twitter": "",
      "talks": [
        {
          "title": "Fin.ai in GTM",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.0024796198,
        0.022934765,
        -0.0042817886,
        -0.022956477,
        -0.058465205,
        -0.04956695,
        -0.058618203,
        0.07715394,
        0.050796308,
        -0.24365164,
        -0.073497996,
        -0.041430898,
        -0.021980122,
        0.029320264,
        -0.048741143,
        0.027665827,
        0.05149296,
        -0.061981544,
        -0.08625531,
        -0.034630194,
        -0.09075421,
        -0.044460516,
        0.005108296,
        -0.02721355,
        -0.017887982,
        0.01096834,
        -0.044121485,
        0.069465056,
        -0.11876062,
        0.4623585,
        -0.07487796,
        0.015251262,
        0.008951619,
        -0.031765927,
        0.0057597994,
        -0.15830067,
        0.024056202,
        -0.0144382985,
        0.03809213,
        0.08291932,
        -0.024134848,
        0.02739878,
        -0.018331943,
        0.023336241,
        -0.01988663,
        -0.049372587,
        -0.086659394,
        -0.0546886,
        0.0047587273,
        0.007988432,
        -0.023114236,
        0.057715144,
        -0.013596893,
        -0.08237412,
        0.04417218,
        0.00092791166,
        0.031704534,
        -0.027494473,
        0.08947827,
        -0.025374334,
        -0.08614586,
        0.022609793,
        0.02270754,
        -0.021066695,
        0.008251749,
        0.007519046,
        0.08961845,
        -0.055514265,
        0.036859684,
        -0.042223867,
        0.030340696,
        0.031265266,
        -0.08070702,
        0.06370511,
        -0.046349857,
        -0.019710919,
        0.03987877,
        0.108450934,
        -0.01317358,
        -0.035643533,
        0.036822874,
        0.06817862,
        -0.06588402,
        -0.056956712,
        0.018443214,
        -0.06678508,
        -0.075138435,
        0.027753219,
        -0.009418968,
        0.07531592,
        0.033298425,
        -0.061936114,
        -0.050551478,
        0.042016264,
        -0.100092426,
        0.048038717,
        -0.016777795,
        0.026577989,
        -0.04028073,
        0.08209306,
        -0.12575297,
        -0.055700082,
        -0.02079577,
        0.010858812,
        -0.007133342,
        -0.12566109,
        -0.028821614,
        -0.034152396,
        0.03326634,
        0.025334124,
        -0.02211232,
        -0.6005473,
        0.014651819,
        -0.068908215,
        -0.0063938745,
        -0.15580304,
        -0.0004090449,
        -0.04829029,
        -0.024868526,
        0.04864242,
        -0.024422046,
        0.005299394,
        0.05712845,
        0.031994447,
        -0.029633014,
        0.10994724,
        0.023824988,
        0.033392034
      ]
    },
    {
      "id": "worldsfair-speaker-366",
      "conference": "worldsfair",
      "name": "Anuj Iravane",
      "role": "AI Lead",
      "company": "Anterior",
      "twitter": "",
      "talks": [
        {
          "title": "Don't be data poor",
          "description": "What do you do when the data you most need to train and evaluate on is the data you're least allowed to keep? It's a bind for anyone building AI in a high-stakes vertical: the cases that would teach your model the most — the rare, the messy, the sensitive — tend to be the ones wrapped in the tightest constraints. In healthcare it's near-absolute. PHI can't be retained, reused, or transformed, so your long-lived datasets can't contain real patient data at all. Synthetic data is the obvious escape hatch, but it has its own trap: synthetic records tend to look synthetic, and a model that passes on fake-looking data tells you nothing about the real thing. So the bar isn't generating data — it's generating data faithful enough to trust. This talk is how we got there. Ask an LLM for a full case in one shot and you get something generic and averaged-out — models are worse at inventing convincing, specific detail than you'd expect. We present our synthetic generation pipeline (and the process around it) that enabled us to create golden datasets at scale. The pipeline features a coarse-to-fine process that enriches a patients medical history layer by layer, with a human in the loop hooks to steer the narrative at each step. You'll leave with ideas on how to build your own synthetic data generation capabilities and how to build a data pipeline your domain experts actually enjoy owning.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.050859362,
        0.051186837,
        -0.03120809,
        -0.029502058,
        0.06254698,
        0.020278001,
        -0.04545469,
        0.055533264,
        0.057482112,
        -0.1968482,
        0.008039814,
        0.0075111645,
        0.012488889,
        -0.07143809,
        -0.059966117,
        -0.042475782,
        0.10069888,
        -0.08915859,
        -0.09114597,
        -0.10704357,
        -0.04633239,
        -0.049491767,
        0.021677801,
        0.053393655,
        -0.022100668,
        0.09572931,
        -0.030284023,
        0.07814818,
        -0.16117969,
        0.41803917,
        -0.11647375,
        -0.053535104,
        0.035489056,
        -0.034677353,
        0.05150261,
        -0.057904366,
        -0.016186314,
        -0.03543846,
        0.004920345,
        0.09900382,
        -0.03333059,
        -0.047161292,
        -0.06193378,
        0.044844463,
        0.017593402,
        0.043019045,
        -0.077709444,
        -0.061591875,
        0.008525214,
        -0.028916478,
        -0.13584745,
        -0.037360463,
        0.0340637,
        0.01889968,
        0.08568543,
        0.091988094,
        -0.011866883,
        -0.10555261,
        0.065567784,
        -0.09697911,
        -0.065858394,
        0.0016952108,
        0.009210464,
        0.022899171,
        -0.01996681,
        -0.054354902,
        0.1427268,
        -0.05628902,
        0.0016960718,
        0.0062038423,
        -0.01774374,
        0.0009558663,
        -0.027514854,
        -0.017339915,
        -0.12611723,
        -0.080025956,
        -0.029444614,
        0.12693341,
        0.020454554,
        -0.0016742512,
        0.03159092,
        0.040808026,
        -0.03961819,
        0.020955672,
        -0.0023101405,
        -0.06670835,
        -0.0442487,
        0.02323453,
        -0.007109229,
        0.09031799,
        0.004715149,
        -0.04253843,
        -0.0415296,
        -0.09322078,
        -0.003490926,
        0.012538345,
        -0.024453795,
        0.03732767,
        0.03749513,
        0.043760393,
        0.097660705,
        0.021294333,
        0.08362691,
        0.007960567,
        -0.06467427,
        -0.07796945,
        0.023216471,
        -0.05566235,
        0.05715361,
        -0.01951088,
        -0.020326193,
        -0.5970737,
        0.06401136,
        0.006057288,
        -0.05702883,
        -0.12355079,
        0.062800914,
        -0.03353246,
        0.026713077,
        0.04068724,
        -0.03628752,
        0.05107627,
        0.012032131,
        0.014652327,
        -0.08373765,
        -0.00092119485,
        0.050416794,
        0.031805113
      ]
    },
    {
      "id": "worldsfair-speaker-367",
      "conference": "worldsfair",
      "name": "Sachin Malhotra",
      "role": "Platform Engineer",
      "company": "Anthropic",
      "twitter": "https://x.com/edorado93",
      "talks": [
        {
          "title": "Give the Agent a Budget, Not a Token",
          "description": "Every agent demo runs with a god-token. Then it ships, and someone has to explain why the helpful AI just rm -rf'd the staging database \"to clean up.\" I run platform infrastructure at a frontier lab, and for the last year my job has partly been: let coding agents do real work against real systems, without ever having to write the postmortem. This talk is the permission model that fell out of that - not RBAC-with-extra-steps, but primitives designed for an actor that's smart, fast, tireless, and occasionally *confidently wrong*. **The four primitives:** - **Asymmetric verbs** - the agent can `quarantine` but not `delete`, `retry` but not `approve`, `propose` but not `merge`. The verb list *is* the security boundary. Stop thinking in resources, start thinking in reversible vs. irreversible actions. - **Regenerating budgets** - every agent identity gets N disruptive actions per window. Burn the budget, you're benched until it refills. No human-in-the-loop until the budget's gone — which means 95% autonomy with a hard ceiling on blast radius. - **The undo test** - if the agent can't undo it, the agent can't do it without a second key. One line, surprisingly load-bearing. - **Tripwires over allow-lists** - let the agent roam, but instrument the three actions that would actually hurt. Cheaper than enumerating everything safe. I'll show the ~200-line policy layer that implements all four, the failure modes each one exists to catch, and the one design I shipped that turned out to be security theater. Tool-agnostic - works whether your agent is touching CI, a database, a cloud account, or your users' files. If you're shipping an agent that does anything more than read, you'll leave with a threat model and a starting policy you can paste into your repo on the flight home.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.052451145,
        0.07686581,
        0.05756742,
        0.021854922,
        0.04249376,
        -0.0024138128,
        -0.097564556,
        -0.022584217,
        0.049317755,
        -0.2477067,
        -0.05687754,
        -0.028751984,
        0.064154565,
        -0.0876944,
        -0.04899178,
        -0.06464964,
        0.07457518,
        -0.085703634,
        -0.048595667,
        -0.0487487,
        -0.026744112,
        -0.08726161,
        0.033236947,
        0.069640405,
        -0.04361274,
        0.124437384,
        0.019707153,
        0.0651388,
        -0.11222141,
        0.44521472,
        -0.07524901,
        0.015194921,
        0.0015695612,
        -0.027703011,
        0.0631322,
        -0.05540563,
        -0.01574798,
        -0.020916717,
        -0.0048839455,
        -0.024311338,
        -0.014164796,
        0.065161034,
        -0.06579224,
        -0.024476536,
        -0.039946314,
        -0.046089746,
        -0.05459181,
        -0.0013226039,
        -0.010214348,
        -0.03240714,
        -0.08663996,
        0.050244104,
        0.070914276,
        0.0033699349,
        0.051739343,
        0.0020968933,
        0.008986253,
        -0.05334792,
        0.14092033,
        -0.020373557,
        -0.0975937,
        -0.009101667,
        0.01777392,
        -0.0010233194,
        -0.03446997,
        -0.09972723,
        0.11418353,
        -0.021401258,
        0.008866087,
        -0.04520407,
        0.016059294,
        0.04837166,
        -0.06058932,
        0.028232371,
        -0.027739959,
        -0.12991159,
        -0.029648636,
        0.07962253,
        -0.0024921147,
        0.030846411,
        -0.016552478,
        0.0858494,
        -0.059752367,
        -0.036278065,
        0.047253214,
        -0.05138241,
        -0.117836915,
        0.011714951,
        0.014732073,
        0.065090336,
        -0.017998114,
        -0.016052255,
        -0.02463391,
        0.04217046,
        -0.058509436,
        0.02944929,
        -0.0483324,
        0.027180485,
        0.04188452,
        0.046282653,
        0.06836027,
        -0.036831815,
        -0.0016320584,
        -0.01951987,
        -0.037692152,
        -0.018598456,
        -0.02656312,
        0.026562784,
        0.044738915,
        -0.01519846,
        -0.0016461792,
        -0.5929941,
        0.023094632,
        -0.014491031,
        -0.06813907,
        -0.1537133,
        0.06591333,
        -0.024425391,
        0.046772845,
        0.07391037,
        -0.0880287,
        0.09214954,
        -0.07419379,
        -0.025115255,
        -0.034415852,
        -0.028641444,
        0.048570983,
        0.0047526876
      ]
    },
    {
      "id": "worldsfair-speaker-368",
      "conference": "worldsfair",
      "name": "Karthik Ranganathan",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Memory Is a Solved Problem. Agent Learning Is Not.",
          "description": "The failures that break multi-agent systems are not reasoning failures, they are handoff failures. One agent works something out and the knowledge dies in its private context, because the only thing that crosses the boundary is output. Memory made each agent better in isolation and changed nothing about what the group knows. The missing primitive is supervised promotion: a deliberate decision about which private learning is worth sharing, moved into common knowledge with the reasoning attached, so trust survives the handoff. Today a human makes that call, and promoted knowledge resolves on read, in any tool, with no retrain or reindex. Those calls are also the training signal for what comes next: orchestrator agents, trained on what matters to the people they serve, that promote on their own. This talk covers how our collective knowledge grew as we approached memory promotion, including what the first build got wrong, and a live look at it working between humans and agents.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.04241831,
        0.07549167,
        -0.0018962539,
        0.07031751,
        0.06269594,
        -0.00026306396,
        -0.06548212,
        -0.036494862,
        0.06349389,
        -0.24956541,
        0.013328281,
        -0.03187285,
        -0.03024777,
        -0.06929054,
        -0.031410858,
        -0.020074844,
        0.10371189,
        -0.06760533,
        -0.07453689,
        -0.067745455,
        0.0075538508,
        -0.09466769,
        -0.027808549,
        0.05089902,
        -0.0050406517,
        0.12974055,
        0.013041824,
        0.08440034,
        -0.05811477,
        0.40253165,
        -0.095857345,
        -0.008533912,
        0.011181485,
        -0.016479714,
        0.08906263,
        -0.07358692,
        0.05330609,
        0.021075726,
        0.080293305,
        0.09447341,
        -0.028940732,
        0.029687269,
        -0.06714204,
        0.07641748,
        0.046755683,
        -0.02072595,
        -0.016130017,
        0.039480787,
        -0.045767803,
        -0.057140462,
        -0.08238493,
        0.007188545,
        0.084314585,
        -0.05648319,
        0.0020344013,
        0.05937119,
        0.004270661,
        -0.023895837,
        0.1047355,
        0.028676314,
        -0.08953808,
        -0.0015447361,
        0.00789522,
        -0.0502968,
        -0.050411895,
        -0.04396327,
        0.08219929,
        -0.053761296,
        0.0170267,
        0.029090779,
        0.018744618,
        0.013990382,
        -0.077407904,
        0.041034147,
        -0.09289399,
        -0.11810278,
        -0.12928452,
        0.08643506,
        -0.033025794,
        0.041317794,
        -0.047524676,
        0.041331425,
        -0.050796587,
        -0.00828226,
        0.011643651,
        -0.00026053732,
        -0.06597193,
        0.027529657,
        -0.08145051,
        0.030783366,
        0.02740106,
        0.019424275,
        0.009879602,
        -0.05176867,
        -0.04777497,
        0.02779546,
        -0.009696457,
        0.076274745,
        0.037769035,
        0.02649838,
        -0.016537845,
        -0.022297638,
        0.024340779,
        -0.027577987,
        0.031159924,
        -0.028238507,
        -0.034083292,
        -0.017912412,
        0.043884054,
        -0.017377663,
        -0.007937495,
        -0.6054834,
        -0.008787667,
        0.031685445,
        -0.04287551,
        -0.14609402,
        0.16374211,
        -0.031941444,
        0.03649248,
        0.06231593,
        -0.014721828,
        0.05233477,
        0.08254881,
        -0.022345189,
        -0.077230185,
        0.040330283,
        0.103545666,
        0.033713162
      ]
    },
    {
      "id": "worldsfair-speaker-369",
      "conference": "worldsfair",
      "name": "Heather Downing",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Agent Memory Is a Solved Problem. Agent Learning Is Not.",
          "description": "The failures that break multi-agent systems are not reasoning failures, they are handoff failures. One agent works something out and the knowledge dies in its private context, because the only thing that crosses the boundary is output. Memory made each agent better in isolation and changed nothing about what the group knows. The missing primitive is supervised promotion: a deliberate decision about which private learning is worth sharing, moved into common knowledge with the reasoning attached, so trust survives the handoff. Today a human makes that call, and promoted knowledge resolves on read, in any tool, with no retrain or reindex. Those calls are also the training signal for what comes next: orchestrator agents, trained on what matters to the people they serve, that promote on their own. This talk covers how our collective knowledge grew as we approached memory promotion, including what the first build got wrong, and a live look at it working between humans and agents.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.028348383,
        0.064022556,
        0.013033279,
        0.045802414,
        0.028616715,
        -0.0040048235,
        -0.05801787,
        -0.06194257,
        0.05924192,
        -0.24077922,
        -0.023298336,
        -0.055137653,
        -0.017879782,
        -0.074150376,
        -0.03202599,
        0.0031755897,
        0.12834741,
        -0.023098709,
        -0.0811763,
        -0.0607043,
        0.026546823,
        -0.08462067,
        -0.04874109,
        0.072695896,
        -0.020752871,
        0.10554492,
        0.026646359,
        0.11236921,
        -0.0919431,
        0.38881537,
        -0.14026548,
        0.009225454,
        0.027183503,
        0.0028597133,
        0.09916196,
        -0.100238994,
        0.04735434,
        -0.02453835,
        0.06873471,
        0.10921679,
        -0.020618599,
        0.03598235,
        -0.044122484,
        0.10089858,
        -0.018494176,
        -0.06547825,
        -0.0069009964,
        0.02412257,
        -0.03359076,
        -0.038525574,
        -0.0855048,
        0.01148691,
        0.083714336,
        -0.034979783,
        0.015035538,
        0.072130546,
        0.052907538,
        -0.055905223,
        0.117909566,
        0.007208084,
        -0.13821186,
        0.01488983,
        -0.042076092,
        0.019195039,
        0.021615619,
        -0.08411905,
        0.07470466,
        -0.034949925,
        -0.003431783,
        -0.018693466,
        0.036531445,
        0.026355978,
        -0.046915546,
        0.035602264,
        -0.078222245,
        -0.121962905,
        -0.087961294,
        0.038067486,
        0.003508375,
        0.05947717,
        -0.06015039,
        0.08552427,
        -0.01262623,
        -0.052854516,
        0.020258572,
        -0.020136693,
        -0.07165502,
        0.02025263,
        -0.08535356,
        0.027997658,
        0.04317007,
        -0.02015244,
        -0.04468551,
        -0.02403581,
        -0.032255694,
        -0.019544346,
        -0.000083159466,
        0.052359454,
        0.023537112,
        0.021437515,
        -0.024555022,
        -0.0690629,
        0.033586815,
        -0.015880384,
        0.026853394,
        -0.0042653144,
        -0.08859828,
        0.032222565,
        0.07194383,
        -0.020525582,
        -0.018532746,
        -0.6008852,
        -0.01292636,
        0.005033428,
        -0.06396655,
        -0.1560793,
        0.14203958,
        0.040614795,
        0.04812813,
        0.020904278,
        -0.029759811,
        0.059573434,
        0.018847113,
        -0.03003678,
        -0.011293231,
        0.004650701,
        0.08219417,
        -0.0010246272
      ]
    },
    {
      "id": "worldsfair-speaker-370",
      "conference": "worldsfair",
      "name": "Remy Guercio",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "An AI Future Without the Lock-In",
          "description": "Every organization navigating AI adoption faces the same trap: the market moves faster than any procurement cycle, no single vendor leads across model quality, interface, sandbox, and data access for more than a few months at a time, and the obvious answer of consolidating behind one platform trades short-term control for long-term lock-in. This session makes the case that the winning strategy is not picking the best walled garden. It is building a connective layer underneath all of them. Tailscale's Remy Guercio walks through the four components required for transformative AI, why vertically integrated stacks are structurally fragile, and how organizations can maintain visibility and control without betting on a single vendor's continued dominance. The second half of the session covers three new capabilities in Aperture, Tailscale's identity-aware AI gateway: Identity-Aware Universal Data Connectors (Public Alpha), which translate Tailscale network identity into scoped access to internal data sources via MCP and API endpoints; a Responsive Chat UI (Public Alpha) that gives non-technical users a mobile-friendly interface to every LLM configured in Aperture; and Sandbox Support (Private Alpha), bringing ephemeral and persistent compute environments into the same identity model. Attendees leave with a framework for evaluating AI platforms that does not depend on picking a winner, and a concrete path to deploying provider-agnostic AI tooling on infrastructure they already run.",
          "day": "Day 4 — Session Day 3",
          "time": "3:20pm-3:40pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09199856,
        0.076887645,
        0.003973473,
        -0.010199029,
        0.04499236,
        -0.043569494,
        -0.057160392,
        -0.013390895,
        0.0680055,
        -0.24322803,
        0.03708365,
        -0.050263505,
        -0.013552771,
        -0.02010404,
        -0.086548276,
        0.029910749,
        0.014734588,
        -0.0533061,
        -0.08526562,
        -0.013125055,
        -0.10109792,
        -0.1027115,
        0.016111078,
        0.031045621,
        -0.110801145,
        0.07780535,
        -0.035770066,
        -0.011414182,
        -0.13566239,
        0.4111675,
        -0.058353275,
        0.03489365,
        0.033855107,
        -0.079739414,
        0.032682773,
        -0.012726251,
        -0.076473035,
        0.008686523,
        0.04055918,
        0.09626706,
        -0.00016060064,
        0.058402352,
        -0.0134912,
        0.001396665,
        -0.015586613,
        -0.06936437,
        -0.10962078,
        0.01009908,
        -0.039924543,
        -0.02191433,
        -0.07437209,
        0.023891764,
        -0.0038595672,
        0.027412407,
        0.05626969,
        0.048712235,
        -0.028425831,
        -0.039646845,
        0.11336654,
        -0.10557252,
        -0.15302758,
        -0.08016214,
        -0.039264124,
        0.03779695,
        -0.015579338,
        -0.022222692,
        0.03403551,
        -0.056806035,
        0.0022171664,
        -0.07338247,
        0.0524839,
        0.034293838,
        -0.029840486,
        -0.008065082,
        -0.05752663,
        -0.10390392,
        -0.040788904,
        0.14904387,
        -0.029583402,
        0.010553672,
        -0.0023475767,
        0.056810748,
        -0.08164512,
        -0.0112222815,
        0.0032917366,
        -0.018429035,
        -0.07311395,
        0.06976625,
        -0.012004682,
        0.070074566,
        0.02100631,
        -0.038606346,
        -0.01217708,
        0.008365518,
        0.02027861,
        -0.07326584,
        -0.056659978,
        0.07768852,
        -0.0065997266,
        0.033133157,
        0.0046220133,
        -0.074433334,
        0.04492918,
        0.00985757,
        0.019642627,
        -0.039448146,
        -0.02515947,
        0.0023240922,
        0.047120985,
        0.028684236,
        -0.03870671,
        -0.6042124,
        -0.035630155,
        0.06157929,
        -0.052780125,
        -0.13558674,
        0.009368164,
        0.008058337,
        -0.014895624,
        0.103405826,
        -0.038811836,
        0.045468442,
        -0.08856875,
        -0.036945704,
        -0.01174971,
        0.076613456,
        0.04753619,
        -0.024415767
      ]
    },
    {
      "id": "worldsfair-speaker-371",
      "conference": "worldsfair",
      "name": "Philipp Schmid",
      "role": "Staff Engineer",
      "company": "Google DeepMind",
      "twitter": "https://x.com/_philschmid",
      "talks": [
        {
          "title": "Agents Without Code: How Skills, YAML, and Filesystems Replaced Python",
          "description": "Six months ago, building an agent meant writing a Python class with a `while` loop, tool definitions in dicts, manual state management or writing custom python functions. Today, you define an agent in a YAML file, drop a `SKILL.md` into a folder, and deploy. This talk traces the arc from \"Agent in Python\" to \"Agent as filesystem\". You'll learn the same agent built three ways: the hard way (Jan 2025), the simple way (Oct 2025), and the zero-code way (today).",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.045002297,
        -0.0008564955,
        0.0117105665,
        -0.027251577,
        0.028700566,
        -0.06275638,
        -0.060963575,
        0.047045942,
        0.050743796,
        -0.23929828,
        -0.028766306,
        -0.029154887,
        0.039464448,
        -0.04368856,
        0.006430904,
        0.0141820265,
        0.12428379,
        -0.0992053,
        -0.043245174,
        -0.1313866,
        -0.045732863,
        -0.06178676,
        -0.031419437,
        0.052772447,
        -0.049464446,
        0.042717744,
        -0.033103246,
        0.05859283,
        -0.078755684,
        0.4902195,
        -0.13551435,
        -0.020499231,
        -0.0028673664,
        -0.023347085,
        0.012908952,
        0.019579481,
        -0.037254486,
        0.029817186,
        0.02801943,
        0.054389533,
        -0.018108262,
        0.025664276,
        -0.03787666,
        0.000066384426,
        -0.02530573,
        -0.041346714,
        -0.036741715,
        0.080972336,
        0.026613439,
        -0.014561328,
        -0.056980792,
        0.017295705,
        -0.003215799,
        -0.028022218,
        0.034555305,
        0.029154398,
        0.06563515,
        -0.11288922,
        0.0829614,
        -0.04071824,
        -0.099970445,
        -0.0042947535,
        -0.011518488,
        0.023207294,
        0.021495866,
        0.018167127,
        0.063377574,
        -0.09683958,
        -0.031802103,
        -0.0574712,
        0.009453878,
        -0.0008542639,
        -0.054278005,
        0.053782336,
        -0.07776916,
        -0.08849261,
        -0.042193267,
        0.01043673,
        -0.032489307,
        0.021095717,
        -0.032286335,
        0.070773974,
        -0.093588494,
        -0.024079068,
        0.01584495,
        0.05158259,
        -0.047042083,
        0.04537012,
        0.0492004,
        0.046223026,
        -0.005238786,
        0.021938875,
        -0.057867445,
        -0.002014916,
        -0.069833644,
        -0.005285903,
        -0.040671933,
        0.010451283,
        -0.05432544,
        0.059424013,
        -0.01764152,
        0.041788712,
        -0.0029543084,
        -0.039299328,
        -0.013624931,
        -0.052142028,
        0.011394047,
        0.0054022693,
        0.046711173,
        -0.036758475,
        -0.07640494,
        -0.5925264,
        0.006197288,
        0.009762018,
        -0.007882965,
        -0.17600235,
        0.039041672,
        -0.014634538,
        0.07599154,
        0.125188,
        -0.0015080102,
        0.090460256,
        -0.040264186,
        -0.033564307,
        0.03743254,
        0.07270101,
        0.06780786,
        0.0063894223
      ]
    },
    {
      "id": "worldsfair-speaker-372",
      "conference": "worldsfair",
      "name": "Arturo Nereu",
      "role": "AI & Game Development",
      "company": "MongoDB",
      "twitter": "https://x.com/arturonereu",
      "talks": [
        {
          "title": "The Next Game Engine Won't Have a Manual",
          "description": "Game development needs to change for the agent era rather than simply dropping an LLM into existing engines. This talk shows the AI systems behind Veselka, using Claude plus Three.js to turn AI into a practical game-development partner and lower the barrier for people who want to build their dream game.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.03971499,
        0.025761455,
        0.015620341,
        0.049819227,
        0.0088120485,
        -0.022899352,
        -0.038335145,
        -0.037499815,
        0.03126303,
        -0.1943931,
        0.0002656689,
        0.020435954,
        0.011769295,
        -0.034070402,
        -0.044442188,
        -0.01514606,
        0.09641235,
        -0.052243683,
        -0.0075955307,
        -0.08301999,
        -0.045942347,
        -0.108629175,
        -0.08198493,
        0.027944878,
        -0.03627067,
        0.18089934,
        -0.034328815,
        -0.020640764,
        -0.09524592,
        0.4738996,
        -0.106138766,
        -0.020230846,
        -0.019430308,
        -0.08584026,
        0.039337225,
        -0.023445174,
        -0.04002082,
        -0.041069273,
        -0.022383621,
        -0.005108326,
        0.026121685,
        0.059658714,
        -0.07439772,
        0.041886665,
        0.023734022,
        -0.06582895,
        -0.037228443,
        0.019586666,
        0.06665685,
        0.050592422,
        -0.037763365,
        0.008788557,
        0.053781163,
        -0.0035397548,
        0.05456385,
        -0.00013150742,
        0.084173866,
        -0.056964166,
        0.078044936,
        -0.024111222,
        -0.121258184,
        -0.015559845,
        0.02060111,
        0.045556184,
        -0.026799317,
        -0.09378493,
        0.039100368,
        -0.058843207,
        0.008447068,
        -0.092948094,
        -0.039566837,
        0.011521563,
        0.0037633262,
        -0.012467296,
        0.003730252,
        -0.103615224,
        -0.06594496,
        0.060714036,
        -0.011019914,
        0.018476635,
        0.0017830675,
        0.04694674,
        -0.07602857,
        -0.01771476,
        0.04961603,
        0.004100471,
        -0.09843441,
        0.069348775,
        -0.010523934,
        0.10965161,
        0.036591727,
        -0.042700578,
        -0.052276865,
        0.026640305,
        -0.014439165,
        -0.029283809,
        -0.00195434,
        0.034564435,
        -0.0063329353,
        0.0034979067,
        0.0063385624,
        -0.05684324,
        0.033139408,
        -0.034850612,
        0.0388308,
        -0.053126104,
        -0.009050899,
        -0.000783689,
        0.11045065,
        0.010067068,
        -0.016478157,
        -0.6048328,
        -0.0130035775,
        0.00955574,
        0.046894975,
        -0.15898189,
        0.026517602,
        -0.0557313,
        0.030762639,
        0.08122696,
        0.0000028186103,
        0.10731357,
        0.05596402,
        -0.03899142,
        0.052011125,
        0.08854122,
        -0.0061880634,
        0.043789573
      ]
    },
    {
      "id": "worldsfair-speaker-373",
      "conference": "worldsfair",
      "name": "David Levine",
      "role": "Founder & CEO",
      "company": "Kiduna Club",
      "twitter": "https://x.com/motocoaster",
      "talks": [
        {
          "title": "Beyond the Lethal Trifecta: Agentic Commerce on the Open Internet at Machine Speed",
          "description": "For decades, the internet has had protocols for routing, identity, encryption, payments, and commerce between people and organizations. It has never had a native way for autonomous agents to possess authority, accountability, or legal standing. On July 1, 2026 that changes. A little known law will take effect that changes the world as we know it. As AI agents move beyond the enterprise firewall, a new form of commerce is emerging. Agents can already search, negotiate, schedule, purchase, settle payments, and coordinate work across networks. But the moment they begin acting independently on behalf of people, businesses, and online organizations, fundamental questions appear: Who does this agent represent? What authority does it possess? Who is responsible when something goes wrong? How do counterparties know they can trust it? This talk explores the \"Lethal Trifecta\" of agentic systems: access to systems, access to networks, and autonomy. Together they create extraordinary capabilities, but they also expose a missing layer in the architecture of the internet itself. Without identity, accountability, governance, and legal standing, agentic commerce remains trapped inside enterprise walls, limited to productivity gains rather than participation in open markets. On the same day as this conference, a new legal framework takes effect that gives autonomous online organizations a registered legal existence, allowing them to hold assets, enter agreements, govern themselves through software, and operate through fleets of agents. Whether you're building agents, agent platforms, autonomous organizations, payment systems, governance systems, or the next generation of internet infrastructure, this shift has global implications, and you'll be the first to know. We'll examine the emerging trust stack for agentic commerce—identity, authority, governance, settlement, and standing—and explore what happens when agents stop acting merely as tools and begin participating as economic actors on the open internet at machine speed.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.06504506,
        0.022337528,
        -0.016921636,
        -0.0022330603,
        0.031386357,
        -0.02721466,
        -0.07443971,
        0.054739118,
        0.016322723,
        -0.20174977,
        -0.07139648,
        -0.08839475,
        -0.03314479,
        -0.06801759,
        -0.06522966,
        -0.00840813,
        0.0983011,
        -0.054121375,
        -0.0447774,
        -0.070454516,
        -0.05065395,
        -0.06276638,
        -0.022487769,
        0.06267832,
        -0.06287421,
        0.12254577,
        -0.019133754,
        0.10302693,
        -0.035818677,
        0.39531866,
        -0.057524607,
        0.047402572,
        0.036854114,
        0.08555659,
        0.0592304,
        -0.09388241,
        -0.038030535,
        -0.050087977,
        0.03251627,
        0.05009276,
        0.010835928,
        0.023759052,
        0.004680365,
        0.08399984,
        0.030808933,
        -0.056966577,
        -0.08308351,
        -0.0089499215,
        0.030426152,
        -0.018346148,
        -0.046025738,
        -0.024391655,
        0.0689203,
        0.060168523,
        0.093487225,
        0.040653717,
        0.10290856,
        -0.014935217,
        0.11254646,
        -0.04898548,
        -0.076806456,
        0.043367695,
        -0.05589719,
        0.0062394952,
        0.02569793,
        -0.07178835,
        0.027179692,
        -0.07260469,
        0.0087392265,
        -0.048607435,
        0.00074920023,
        0.09341353,
        -0.055556025,
        -0.038183033,
        -0.069024496,
        -0.11219888,
        -0.098290905,
        0.065118805,
        -0.05093599,
        0.024153328,
        -0.011327821,
        0.06534493,
        -0.06571433,
        -0.060123153,
        0.007137976,
        -0.018757394,
        -0.014933875,
        0.045880586,
        -0.005485709,
        0.09324635,
        0.014989317,
        -0.08196931,
        -0.12828478,
        -0.0073084766,
        -0.065322615,
        0.029877061,
        -0.021346262,
        0.013463781,
        -0.010465471,
        -0.06215409,
        0.010294623,
        -0.01790708,
        0.07050413,
        0.019828772,
        0.003941748,
        0.0031665962,
        0.07629234,
        0.06821853,
        0.11280292,
        -0.0035341277,
        -0.0016895315,
        -0.6251624,
        0.019159708,
        0.021101337,
        -0.06043828,
        -0.1334866,
        0.07747701,
        -0.0067023095,
        0.018470552,
        0.07024827,
        -0.03221804,
        0.07083842,
        0.05152473,
        0.0054541873,
        -0.007550706,
        0.018322308,
        0.081438,
        0.044579815
      ]
    },
    {
      "id": "worldsfair-speaker-374",
      "conference": "worldsfair",
      "name": "Sai Krishna Rallabandi",
      "role": "Director, Data Science",
      "company": "Fidelity Investments",
      "twitter": "https://x.com/Saikallis9012",
      "talks": [
        {
          "title": "Wearing the Agent: Engineering a Family-and-Friends Personal Agent, from Group Chats to Glasses",
          "description": "Judith is a personal AI agent that has run in daily production for a year, used by more than a dozen family and friends across WhatsApp group chats, Telegram, and Discord. This talk covers the engineering for a safe multi-tenant personal agent: permissioning, long-lived memory across FAISS + Neo4j + curated notes, scheduled subagents, and message-time guardrails for privacy, recipient safety, and prompt-injection defense. It then shows how the agent moves onto low-cost smart glasses, capturing visual memory, helping with navigation and in-store tasks, and maintaining conversational latency with on-device speech recognition, cloud reasoning, and a custom neural voice. Includes live demos plus practical takeaways on multi-user agent design, durable memory, defensive agent engineering, and wearable ambient interfaces.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.009505526,
        0.06559159,
        -0.05039003,
        0.04816719,
        0.018188428,
        -0.027135398,
        -0.09861665,
        0.00769141,
        0.055876017,
        -0.23017651,
        -0.049679935,
        0.017911702,
        0.013363826,
        -0.007071925,
        -0.024360958,
        0.027153248,
        0.08359702,
        -0.094514236,
        -0.03797572,
        -0.05582061,
        -0.02249576,
        -0.07607933,
        -0.03477177,
        0.014369806,
        -0.01908618,
        0.073981375,
        0.0062508364,
        0.066149734,
        -0.09882149,
        0.46379703,
        -0.085594386,
        0.03810006,
        0.037425343,
        -0.017241834,
        0.06707222,
        -0.06257987,
        -0.032743696,
        0.0025578754,
        -0.005656252,
        0.04893391,
        -0.0014125553,
        0.05407405,
        -0.0007820211,
        0.043991286,
        0.013167532,
        -0.023075819,
        -0.05790121,
        0.05854638,
        -0.03153593,
        -0.012843559,
        -0.05991747,
        -0.022541853,
        0.099563494,
        -0.016849305,
        0.019586885,
        0.03873512,
        0.010338187,
        -0.09196583,
        0.074372075,
        -0.062709145,
        -0.108773135,
        0.018452242,
        0.06764437,
        -0.013248225,
        0.042961176,
        -0.12898035,
        0.06729334,
        -0.02370408,
        -0.029179076,
        -0.01983323,
        -0.005164383,
        -0.025054269,
        -0.08482266,
        0.04966125,
        -0.07043467,
        -0.13897,
        -0.035719633,
        0.081718914,
        -0.068970226,
        0.034632508,
        0.079094544,
        0.07449726,
        -0.056460552,
        -0.037324727,
        0.004405437,
        -0.043434046,
        -0.03595591,
        0.03148425,
        0.015910128,
        0.037236977,
        0.007349776,
        -0.041281868,
        -0.04978803,
        -0.049326386,
        -0.06969381,
        0.021841541,
        -0.076051936,
        0.043195914,
        -0.015267888,
        0.042792793,
        -0.021127155,
        -0.0032679636,
        0.038283177,
        0.023424456,
        -0.010111142,
        -0.04202769,
        0.018630907,
        -0.010806273,
        0.097752824,
        0.06650448,
        -0.027182385,
        -0.5962066,
        -0.029468654,
        -0.055911016,
        -0.0129069695,
        -0.1523302,
        0.033008404,
        -0.07047962,
        0.069804065,
        0.057550374,
        -0.05281764,
        0.12223505,
        0.0034647812,
        0.009160337,
        -0.04938145,
        0.05305772,
        0.069335006,
        -0.085826226
      ]
    },
    {
      "id": "worldsfair-speaker-375",
      "conference": "worldsfair",
      "name": "Ara Khan",
      "role": "Founder",
      "company": "Cline",
      "twitter": "https://x.com/arafatkatze",
      "talks": [
        {
          "title": "Making agents you could never make",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.0052299597,
        0.088144444,
        -0.0051731654,
        -0.039482147,
        0.03320788,
        -0.0062230583,
        -0.05113093,
        0.041947525,
        -0.011405764,
        -0.2355254,
        0.026122084,
        -0.09453415,
        0.0095634805,
        -0.033796072,
        -0.05487134,
        -0.011606756,
        0.13091327,
        -0.11352355,
        -0.08884092,
        -0.045207124,
        -0.033101063,
        -0.108581886,
        -0.02009792,
        -0.010096109,
        -0.07290292,
        0.09155318,
        -0.08766378,
        0.058619052,
        -0.110776305,
        0.43796784,
        -0.08118514,
        -0.017585197,
        0.08881582,
        0.029587852,
        0.06195789,
        -0.07763832,
        0.02576246,
        -0.025636252,
        0.04627364,
        0.057741735,
        -0.009851535,
        0.05550943,
        -0.009476212,
        0.019143302,
        -0.046065435,
        0.044115312,
        -0.07197978,
        -0.008847794,
        -0.032395992,
        0.020647192,
        -0.052785903,
        0.013406674,
        0.00060956506,
        -0.025295239,
        0.04339855,
        -0.02187744,
        -0.0058929543,
        -0.11015867,
        0.003918103,
        -0.07195624,
        -0.08280216,
        0.022484764,
        -0.014654345,
        0.01823096,
        -0.023254886,
        -0.061458223,
        0.074050285,
        -0.11500687,
        0.02627399,
        -0.06170278,
        0.014557322,
        -0.023882968,
        -0.025855772,
        0.056034625,
        -0.10885254,
        -0.062708616,
        -0.029215215,
        0.055460233,
        -0.011233821,
        0.040562984,
        0.05052979,
        0.026662579,
        -0.120218635,
        -0.07931608,
        -0.01998776,
        -0.08551407,
        -0.083096035,
        0.13984221,
        -0.02258316,
        0.1209691,
        0.053144757,
        0.018317267,
        -0.038016427,
        -0.02090919,
        -0.03381895,
        0.005635756,
        -0.11143761,
        0.046574846,
        -0.04295515,
        0.03774788,
        0.042942632,
        -0.044039432,
        0.04460196,
        0.0004311147,
        -0.004264361,
        0.0014338682,
        0.026385972,
        -0.03452363,
        0.06079192,
        0.01586515,
        -0.06205544,
        -0.5886451,
        0.058429338,
        0.02156298,
        -0.0079612965,
        -0.11407701,
        -0.001885235,
        -0.034284018,
        0.0020721683,
        0.08450604,
        -0.08367177,
        0.062314723,
        -0.010733087,
        0.034161672,
        -0.065034695,
        0.00982271,
        0.020643832,
        0.007630089
      ]
    },
    {
      "id": "worldsfair-speaker-376",
      "conference": "worldsfair",
      "name": "Subbiah Sethuraman",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Why We Killed Our Multi-Agent Pipeline: Lessons From Pharma Commercial Intelligence",
          "description": "Key takeaways: A practical design principle for agentic systems in regulated, high-stakes domains: derive the architecture from agent behavior, don't impose it. Concrete patterns the audience can apply this week — domain knowledge graphs as agent context, deterministic preprocessing as a complement to agentic reasoning, reference-based context management. An honest case study from production: what worked, what didn't, and the open architectural questions we're still working on. Abstract : We lead the architecture and AI engineering org behind ZS Associates' commercial intelligence platform for pharmaceutical brand teams. The product has two surfaces: a proactive alert system that delivers signal-driven intelligence packets when a brand's KPIs move, and a conversational analytics chat where business users ask ad-hoc questions. A year ago we built both surfaces as separate V1 stacks. They broke in different ways. The diagnosis was the same: we had decided on the structure before we knew what the agent actually needed. This talk is about the design principle that came out of rebuilding both — and what it produced. The architecture is derived, not designed. We stopped trying to predict what scaffolding the agent would need and started designing the system around what the agent's behavior, on real production tasks, actually demanded. Tools, context, structure, and guardrails get introduced at the points where the agent's reasoning needs them — and nowhere else. What that produced is an architecture that's smaller than V1, not bigger. A single agent owns each investigation end-to-end across both surfaces, launching parallel sub-agents when the work needs them — not according to a pre-defined topology. A pharmaceutical commercial knowledge graph — HCPs, accounts, payers, territories, brands, KPIs and the relationships between them — gives the agent the domain context it needs without prompt-engineering heroics. Statistical signal detection runs deterministically before the agent wakes up, so the agent's job is to explain signals, not find them. Raw query results stay out of the context window through a reference-pattern that lets the agent reason over data without drowning in it. Each of those decisions came from watching an agent struggle on a real task and asking what does it need here? — not from sketching the architecture in a doc and forcing the agent into it. The patterns generalize. If you're shipping agents over messy enterprise data — finance, supply chain, claims, operations — the failure modes and the fixes will look familiar. We'll close with the open questions and the pieces we haven't solved yet.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        -0.001373472,
        0.048860114,
        0.023720635,
        0.029443882,
        0.015180563,
        -0.04180258,
        -0.11503615,
        0.024453709,
        0.05467775,
        -0.22676885,
        0.019296415,
        -0.039604172,
        0.001829844,
        -0.10059228,
        -0.070139594,
        -0.008196965,
        0.08652601,
        -0.110944994,
        -0.07417497,
        -0.04700693,
        -0.024871897,
        -0.08565062,
        -0.00498906,
        0.07797074,
        -0.06173701,
        0.04455189,
        -0.04456474,
        0.0632075,
        -0.055059522,
        0.4611899,
        -0.06696395,
        0.0035037068,
        -0.014050928,
        -0.00034271105,
        0.059525765,
        -0.08733246,
        -0.008428146,
        -0.0478604,
        0.042693403,
        0.044751305,
        0.008056197,
        0.065428756,
        -0.024983872,
        0.048133805,
        -0.056417443,
        -0.0308628,
        -0.021836195,
        0.039633244,
        -0.065458246,
        -0.017532399,
        -0.06759242,
        -0.07576231,
        0.057959218,
        -0.020130213,
        -0.0044688247,
        0.020303193,
        0.00001868878,
        0.009183958,
        0.1588235,
        -0.008582134,
        -0.097543664,
        0.0013356621,
        0.0026919409,
        -0.028932212,
        -0.013213696,
        -0.08240601,
        0.08087769,
        -0.05390448,
        0.03833187,
        -0.026194656,
        -0.014942758,
        -0.010184344,
        -0.131003,
        -0.02508625,
        -0.055794306,
        -0.10313211,
        -0.07274747,
        0.09323845,
        -0.006647853,
        0.065580405,
        -0.023006741,
        0.08495468,
        -0.055228207,
        -0.0680804,
        0.049890712,
        -0.02927208,
        -0.029490354,
        0.05948656,
        -0.01631265,
        0.08876924,
        0.037201878,
        0.01726666,
        -0.04936751,
        -0.089741506,
        -0.01377392,
        0.025281401,
        -0.044632614,
        0.045096975,
        -0.022409474,
        0.03389931,
        -0.035521615,
        -0.047379643,
        0.0825222,
        -0.015414736,
        0.043546826,
        -0.07924029,
        0.014014845,
        -0.018554494,
        0.08556969,
        0.052396357,
        -0.020582236,
        -0.600551,
        -0.008719877,
        -0.016336126,
        0.00878464,
        -0.15172018,
        0.07886174,
        -0.02467938,
        0.053948794,
        0.02666112,
        -0.028216638,
        0.010406608,
        -0.008774781,
        -0.007388201,
        -0.049130093,
        0.019758122,
        0.049952377,
        0.0071510803
      ]
    },
    {
      "id": "worldsfair-speaker-377",
      "conference": "worldsfair",
      "name": "Abhilash Asokan",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Why We Killed Our Multi-Agent Pipeline: Lessons From Pharma Commercial Intelligence",
          "description": "Key takeaways: A practical design principle for agentic systems in regulated, high-stakes domains: derive the architecture from agent behavior, don't impose it. Concrete patterns the audience can apply this week — domain knowledge graphs as agent context, deterministic preprocessing as a complement to agentic reasoning, reference-based context management. An honest case study from production: what worked, what didn't, and the open architectural questions we're still working on. Abstract : We lead the architecture and AI engineering org behind ZS Associates' commercial intelligence platform for pharmaceutical brand teams. The product has two surfaces: a proactive alert system that delivers signal-driven intelligence packets when a brand's KPIs move, and a conversational analytics chat where business users ask ad-hoc questions. A year ago we built both surfaces as separate V1 stacks. They broke in different ways. The diagnosis was the same: we had decided on the structure before we knew what the agent actually needed. This talk is about the design principle that came out of rebuilding both — and what it produced. The architecture is derived, not designed. We stopped trying to predict what scaffolding the agent would need and started designing the system around what the agent's behavior, on real production tasks, actually demanded. Tools, context, structure, and guardrails get introduced at the points where the agent's reasoning needs them — and nowhere else. What that produced is an architecture that's smaller than V1, not bigger. A single agent owns each investigation end-to-end across both surfaces, launching parallel sub-agents when the work needs them — not according to a pre-defined topology. A pharmaceutical commercial knowledge graph — HCPs, accounts, payers, territories, brands, KPIs and the relationships between them — gives the agent the domain context it needs without prompt-engineering heroics. Statistical signal detection runs deterministically before the agent wakes up, so the agent's job is to explain signals, not find them. Raw query results stay out of the context window through a reference-pattern that lets the agent reason over data without drowning in it. Each of those decisions came from watching an agent struggle on a real task and asking what does it need here? — not from sketching the architecture in a doc and forcing the agent into it. The patterns generalize. If you're shipping agents over messy enterprise data — finance, supply chain, claims, operations — the failure modes and the fixes will look familiar. We'll close with the open questions and the pieces we haven't solved yet.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "sponsor"
        }
      ],
      "embedding": [
        0.029929304,
        0.05207936,
        0.05640457,
        0.02907361,
        0.041947313,
        -0.039874703,
        -0.045281902,
        0.0030028364,
        0.038945276,
        -0.20041136,
        0.0016510034,
        -0.029089123,
        0.008827937,
        -0.07377293,
        -0.070380285,
        -0.024854168,
        0.11675351,
        -0.13913395,
        -0.053366102,
        -0.05699224,
        -0.07912905,
        -0.07292582,
        -0.06775355,
        0.052174203,
        -0.06819398,
        0.04621088,
        -0.0672822,
        0.05503536,
        -0.08596368,
        0.44887263,
        -0.06782987,
        -0.007612668,
        0.008851449,
        -0.010706352,
        0.045900956,
        -0.0967462,
        -0.022454752,
        -0.013272504,
        0.009843063,
        0.017427785,
        0.019806078,
        0.016907392,
        -0.02963708,
        0.06984396,
        -0.08638099,
        -0.022743648,
        0.018894942,
        0.00797415,
        -0.049451377,
        -0.04039109,
        -0.09536551,
        -0.063955225,
        0.0087298965,
        -0.039277624,
        -0.024967616,
        0.054839715,
        -0.0151231345,
        -0.05088279,
        0.10841713,
        -0.0015244728,
        -0.046388205,
        -0.052571263,
        0.022972202,
        0.010256421,
        -0.014369063,
        -0.10034697,
        0.12299479,
        -0.022601286,
        0.005286425,
        -0.0035025936,
        -0.018458799,
        0.034134604,
        -0.12598987,
        -0.022267606,
        -0.039301462,
        -0.06983991,
        -0.06278358,
        0.10802719,
        0.016716313,
        0.045043696,
        -0.0408612,
        0.06959029,
        -0.009480206,
        -0.04021703,
        0.03980679,
        -0.014169164,
        -0.021462178,
        0.06701494,
        0.0063668024,
        0.06139092,
        0.0103994645,
        0.012835258,
        -0.047634516,
        -0.11149421,
        -0.076966956,
        0.04183832,
        -0.0085009085,
        0.10028058,
        -0.01942629,
        0.052214067,
        -0.019503914,
        -0.03244419,
        0.08201473,
        -0.008843471,
        0.014583851,
        -0.05812081,
        0.004042343,
        -0.03190027,
        0.083923265,
        -0.012731055,
        -0.026796486,
        -0.625679,
        -0.022675466,
        -0.014481665,
        -0.054219477,
        -0.16142103,
        0.07831748,
        0.004527441,
        0.0113036595,
        0.049235497,
        -0.03223098,
        0.021855738,
        -0.004580868,
        0.018705374,
        -0.0061754724,
        -0.011593206,
        0.03477388,
        0.02229104
      ]
    },
    {
      "id": "worldsfair-speaker-378",
      "conference": "worldsfair",
      "name": "Aliisa Rosenthal",
      "role": "General Partner",
      "company": "Acrew Capital",
      "twitter": "",
      "talks": [
        {
          "title": "Reverse-Engineering the AI Buyer",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        -0.050986778,
        0.054124,
        -0.04788405,
        0.023755211,
        0.027597908,
        0.01021205,
        -0.0022881273,
        -0.01636377,
        0.02618205,
        -0.17062534,
        -0.005651102,
        -0.109218515,
        0.052091554,
        -0.08380138,
        -0.09848022,
        0.023398308,
        0.055748455,
        -0.027704984,
        -0.06261406,
        -0.12877876,
        0.022101207,
        -0.04770289,
        -0.0017219167,
        0.044906847,
        -0.03750181,
        0.054538436,
        0.027129399,
        0.035124514,
        -0.087474525,
        0.43070248,
        -0.106047526,
        0.021535838,
        -0.0044351635,
        0.08594128,
        0.04766121,
        -0.03055144,
        0.016934516,
        -0.03421165,
        0.0033277199,
        0.08392595,
        -0.01238702,
        0.04197213,
        0.030861953,
        0.13330393,
        -0.081254,
        -0.09095112,
        -0.06812523,
        -0.10198717,
        -0.030001385,
        -0.014935733,
        0.030551169,
        -0.027871631,
        -0.012893271,
        -0.09107132,
        0.029462716,
        0.06702564,
        0.008306212,
        -0.08802081,
        0.09990281,
        0.0017397017,
        -0.16493665,
        0.039416507,
        0.010235438,
        0.0563807,
        0.057484537,
        -0.023280859,
        0.09490686,
        0.013324727,
        0.019964585,
        -0.0844317,
        0.011964222,
        0.079053685,
        -0.062261354,
        0.027011022,
        -0.123165384,
        -0.053808272,
        -0.0014040597,
        0.04223647,
        0.017247528,
        -0.015493982,
        0.038827002,
        0.03206193,
        -0.10415625,
        -0.054031193,
        -0.016811928,
        -0.040163003,
        -0.02195105,
        0.07718746,
        0.02929573,
        0.020084297,
        0.060776748,
        -0.036374398,
        -0.077884965,
        0.0104890205,
        -0.121225394,
        0.049675528,
        -0.096320316,
        0.0463734,
        0.022540934,
        0.021251276,
        -0.014882609,
        -0.063568085,
        0.078396946,
        -0.014511424,
        -0.020621551,
        0.02324126,
        -0.010737113,
        -0.020626543,
        0.045153204,
        -0.07995706,
        -0.042737555,
        -0.57616615,
        -0.0015823882,
        -0.027741155,
        -0.09321842,
        -0.15131876,
        -0.0056504505,
        -0.04535702,
        0.06745826,
        0.11018824,
        -0.08523813,
        -0.010480952,
        -0.021524714,
        0.03615143,
        0.09319593,
        0.03878366,
        0.09325047,
        0.00825761
      ]
    },
    {
      "id": "worldsfair-speaker-379",
      "conference": "worldsfair",
      "name": "Christopher Lovejoy",
      "role": "Member of Technical Staff",
      "company": "Anthropic",
      "twitter": "https://x.com/ChrisLovejoy_",
      "talks": [
        {
          "title": "AI Benchmarks for Vertical Industries: Why we're not measuring what we need to and how to unlock real-world ROI",
          "description": "AI is acing the tests we set for it. So why are so many production deployments falling flat? This talk draws on lessons from building Anterior's internal benchmark for real-world healthcare tasks and how to translate real-world performance into concrete measurement rubrics, use imperfect synthetic data, avoid common pitfalls, and apply the approach to any vertical domain.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.09526255,
        0.048192903,
        -0.032440014,
        0.0081168,
        0.038086735,
        -0.015199478,
        -0.077612445,
        0.038977828,
        0.04134661,
        -0.20531438,
        -0.06758833,
        -0.055173196,
        0.06533278,
        -0.056939736,
        -0.030490011,
        0.069328025,
        0.10895884,
        -0.032613233,
        -0.10414034,
        -0.0439365,
        -0.012527876,
        -0.049874548,
        0.0143455,
        0.070128575,
        -0.09468292,
        0.07152271,
        0.02821082,
        0.038367536,
        -0.13883653,
        0.42496648,
        -0.13866624,
        -0.0046958765,
        -0.07115047,
        0.003892193,
        0.06657997,
        -0.08158779,
        0.021669304,
        0.0013052825,
        0.03085079,
        0.025087176,
        0.04065072,
        0.026286058,
        -0.03533577,
        -0.010415188,
        0.04173026,
        -0.0332912,
        -0.05525652,
        -0.047494642,
        0.0051701413,
        -0.10477401,
        -0.15133865,
        -0.006479135,
        0.019293344,
        0.008281285,
        0.07831769,
        0.06710279,
        0.015007386,
        -0.05459848,
        0.15018137,
        -0.00935095,
        -0.068531185,
        -0.023694608,
        0.018158644,
        0.030372424,
        0.029953929,
        -0.061968308,
        0.0651923,
        -0.024713738,
        -0.01613637,
        -0.061351854,
        0.01599875,
        0.041553263,
        -0.11817031,
        -0.031471636,
        -0.06342508,
        -0.11047311,
        -0.09074839,
        0.049386553,
        -0.044699162,
        0.02669135,
        0.013415091,
        0.09397962,
        0.012451679,
        0.00036683647,
        0.008613416,
        -0.0062019643,
        -0.043771964,
        0.015167347,
        -0.008511671,
        0.07675182,
        -0.007544334,
        -0.13728231,
        -0.0073107914,
        -0.041101635,
        -0.033299603,
        -0.013923712,
        -0.014357852,
        0.038206834,
        0.009189257,
        0.039201573,
        -0.03469202,
        0.0061407024,
        0.080955304,
        -0.044181805,
        0.013365126,
        0.008670387,
        -0.0010519954,
        -0.011077452,
        0.06642167,
        -0.00042707013,
        -0.039987218,
        -0.6132815,
        0.023137905,
        0.021753566,
        -0.028810034,
        -0.15152414,
        0.0562412,
        0.012832811,
        -0.01620317,
        0.060211815,
        -0.04834305,
        0.014737106,
        0.017888516,
        -0.012775135,
        -0.020534959,
        0.054132253,
        0.057528682,
        0.01905515
      ]
    },
    {
      "id": "worldsfair-speaker-380",
      "conference": "worldsfair",
      "name": "Byung-Gon (Gon) Chun",
      "role": "Founder & CEO",
      "company": "FriendliAI",
      "twitter": "",
      "talks": [
        {
          "title": "The Frontier AI Inference Cloud for Agents",
          "description": "Agents have changed the economics of AI inference. A chatbot's cost scales roughly linearly with the number of requests; an agent's scales multiplicatively. A single task can fan out into hundreds of model calls, each carrying a repeated context prefix and adding latency that compounds across tool calls and reasoning steps.\n\nAs open-weight models keep improving and agentic workloads grow, this shift exposes the limits of traditional request-level optimization. Inference infrastructure becomes a first-class concern, one that often shapes performance and cost as much as the model itself.\n\nIn this talk, we explore what changes when you optimize for the whole task rather than the individual request, and how FriendliAI is rethinking the inference cloud for the era of agentic AI.\n\nDescription: Agents' inference costs scale multiplicatively. Optimize the whole task, not the request. See how FriendliAI builds the inference cloud for agents.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.07817,
        0.07553309,
        -0.034608554,
        -0.0015344805,
        0.02713866,
        -0.0539121,
        -0.087030895,
        -0.018300466,
        0.070302635,
        -0.23726471,
        -0.07233995,
        -0.05813441,
        0.0023698297,
        -0.048959978,
        -0.08007695,
        0.06426985,
        0.061357018,
        -0.061489385,
        -0.08450947,
        -0.018430002,
        -0.06537176,
        -0.054625697,
        -0.015517432,
        0.018733643,
        0.017895065,
        0.08336856,
        0.0012176452,
        0.07875155,
        -0.14655401,
        0.39271793,
        -0.07789653,
        -0.0062253787,
        0.04233306,
        -0.02105218,
        0.031649247,
        -0.05330706,
        0.012492308,
        -0.030902674,
        -0.014036323,
        0.03904579,
        0.0344906,
        0.09850697,
        -0.030170774,
        0.062284473,
        -0.033360936,
        -0.04458975,
        -0.081208184,
        0.024432208,
        -0.0228167,
        0.036989298,
        -0.08102201,
        0.02149832,
        0.056617003,
        0.03785293,
        0.04224014,
        -0.0006011238,
        0.026048915,
        -0.14614356,
        0.10667641,
        -0.034300134,
        -0.1594544,
        0.010199388,
        -0.007861906,
        0.05149812,
        -0.03936316,
        -0.052027266,
        -0.026491828,
        -0.07153293,
        0.009107739,
        -0.08669975,
        -0.00015914296,
        0.028173521,
        -0.050289117,
        -0.020214193,
        -0.037162185,
        -0.115103185,
        -0.02843084,
        0.09078432,
        -0.01164832,
        0.00036862085,
        -0.041814137,
        0.076498166,
        -0.04573585,
        -0.083422504,
        0.057095565,
        -0.04228963,
        -0.0714675,
        0.009048004,
        -0.029887917,
        0.06785048,
        0.06142817,
        -0.003942754,
        -0.065570354,
        0.03541655,
        -0.07914724,
        0.046854474,
        -0.047787756,
        0.0025572882,
        -0.004120154,
        0.07027083,
        0.015648184,
        0.022636982,
        0.053715616,
        0.0038038848,
        -0.007745231,
        -0.047380626,
        0.06925671,
        -0.008552148,
        0.020940486,
        0.015496236,
        -0.04973289,
        -0.62818223,
        -0.012224838,
        -0.0031794556,
        -0.0436412,
        -0.13289066,
        -0.0027558808,
        -0.06819542,
        0.056777723,
        0.074901946,
        -0.04549957,
        0.07742079,
        0.047045916,
        0.047073103,
        -0.02064803,
        0.019250555,
        0.03872784,
        0.046406575
      ]
    },
    {
      "id": "worldsfair-speaker-381",
      "conference": "worldsfair",
      "name": "Lena Hall",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "The Signal Layer: What to Build When Anything Can Be Built",
          "description": "AI has made implementation faster, cheaper, and more widely available. That changes the real bottleneck in software.\n\nWhen every team can generate code, spin up agents, prototype workflows, and ship demos faster than ever, the advantage moves to a different layer: knowing what is worth building, who it is for, how people will discover it, and how the product should behave once they do.\n\nThis talk introduces the Signal Layer: the system of public signals, user intent, agent experience, distribution loops, and product judgment that helps builders decide what deserves to exist before they commit time, infrastructure, and trust to building it.\n\nWe will look at how AI changes the software lifecycle from \"can we build it?\" to \"should this exist?\" and how developers, AI engineers, and technical leaders can design products that earn adoption instead of producing impressive demos that disappear.\n\nWhen anything can be built, the most valuable builders are the ones who can read signal early, shape the right experience, and build the thing users were already moving toward.",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.027627833,
        0.026266402,
        0.03624842,
        -0.002919029,
        0.025156097,
        -0.03802796,
        0.0048410688,
        0.032207765,
        0.021618512,
        -0.23456046,
        0.010719343,
        -0.028543526,
        -0.028293004,
        -0.0974726,
        -0.05115359,
        0.051962063,
        0.11973917,
        -0.06269497,
        -0.07107899,
        -0.07506742,
        -0.034275394,
        -0.091803335,
        -0.0030174407,
        0.061632074,
        -0.017653344,
        0.09504229,
        -0.0015824455,
        0.031080034,
        -0.09633669,
        0.41222587,
        -0.051443648,
        0.07128971,
        -0.015650993,
        -0.012866796,
        0.061371442,
        -0.04028814,
        -0.019403668,
        -0.0048956643,
        0.040035404,
        0.0656878,
        0.055687968,
        0.07338044,
        -0.002244458,
        -0.0006090033,
        -0.013623178,
        -0.031573232,
        -0.10554843,
        -0.055830803,
        0.0002478726,
        0.013750723,
        -0.005804947,
        0.008572875,
        0.02502486,
        -0.03605167,
        0.03098693,
        0.048625987,
        -0.03610441,
        -0.05411007,
        0.11623923,
        -0.025723351,
        -0.1319929,
        -0.012955691,
        -0.07012316,
        0.041431423,
        0.000058097314,
        -0.09811847,
        0.09012038,
        -0.010207326,
        -0.028686533,
        -0.115055434,
        0.011505406,
        0.017703904,
        -0.055373408,
        0.02631579,
        -0.053818285,
        -0.10316998,
        -0.037622064,
        0.05394048,
        -0.006425355,
        0.015582703,
        -0.021770388,
        0.05401785,
        -0.09211374,
        -0.024687912,
        -0.0028838534,
        0.02773988,
        -0.09290604,
        0.050687227,
        -0.011110487,
        0.09731892,
        0.0021559792,
        0.030840676,
        -0.05921242,
        0.03441025,
        -0.0025671604,
        0.04746115,
        -0.07081976,
        -0.0036274288,
        -0.04038733,
        0.09205177,
        -0.06697434,
        -0.08789654,
        -0.014181507,
        -0.016373979,
        -0.015343136,
        -0.04779258,
        -0.02273048,
        0.014115528,
        0.06402436,
        -0.010039835,
        -0.013026651,
        -0.640343,
        0.024846008,
        0.030971143,
        -0.009065241,
        -0.14147818,
        0.0117570935,
        0.007922004,
        -0.022128122,
        0.0969655,
        -0.043188795,
        0.013137833,
        -0.048488114,
        -0.068422854,
        0.05399897,
        0.012381712,
        0.09218218,
        0.064125575
      ]
    },
    {
      "id": "worldsfair-speaker-382",
      "conference": "worldsfair",
      "name": "Rémi Louf",
      "role": "",
      "company": ".txt",
      "twitter": "https://x.com/remilouf",
      "talks": [
        {
          "title": "Agent Frameworks Considered Harmful",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "3:45pm-4:05pm",
          "type": "session"
        }
      ],
      "embedding": [
        0.05512697,
        -0.06569905,
        0.014373986,
        -0.02464952,
        0.058736008,
        -0.069009915,
        -0.02342617,
        0.04187092,
        0.025511842,
        -0.2099918,
        -0.059428133,
        -0.025379507,
        0.0052410862,
        -0.017109023,
        -0.090461746,
        -0.01634428,
        0.11058901,
        -0.071850725,
        -0.057588577,
        -0.06255052,
        -0.0952094,
        -0.13007729,
        -0.029641682,
        -0.00003280011,
        -0.017612787,
        0.12639634,
        -0.04018799,
        0.043068897,
        -0.13634047,
        0.4683614,
        -0.09799532,
        0.034228068,
        0.01980561,
        -0.029636508,
        0.008024343,
        -0.041183945,
        0.0014334589,
        0.018924437,
        0.029533107,
        0.062499214,
        0.0015822238,
        0.010697574,
        -0.016137654,
        -0.02898746,
        0.016039835,
        0.024485698,
        -0.08421231,
        -0.0435744,
        -0.043928552,
        -0.032530915,
        -0.086022705,
        0.08407557,
        -0.010554594,
        -0.06865492,
        -0.008469122,
        -0.027422948,
        0.040114246,
        -0.04156508,
        0.056530993,
        -0.021858145,
        -0.033415336,
        -0.13660896,
        -0.04142196,
        0.016174585,
        0.0126376,
        -0.00048521656,
        0.06519574,
        -0.071845815,
        0.030862577,
        -0.07360062,
        0.06008095,
        -0.021860702,
        -0.011129799,
        0.0319242,
        -0.054384593,
        0.023134515,
        -0.046017494,
        0.09731842,
        -0.033124954,
        0.027345942,
        -0.014871384,
        0.052501906,
        -0.10688537,
        -0.03021575,
        0.033279084,
        -0.05958357,
        -0.111933306,
        0.027204284,
        0.104960404,
        0.051369328,
        0.058150783,
        -0.018097663,
        -0.012606369,
        0.0257255,
        -0.05533136,
        -0.0033444678,
        -0.062353354,
        0.07288177,
        0.041504603,
        0.028659867,
        -0.07411124,
        -0.017428515,
        0.06435998,
        -0.08779612,
        -0.037011407,
        -0.06765528,
        -0.07959346,
        -0.008284726,
        0.0031394775,
        0.0010555089,
        -0.018160451,
        -0.5771468,
        -0.06927892,
        -0.011379818,
        0.039305806,
        -0.17028551,
        0.044388324,
        0.0010536897,
        -0.0083900625,
        0.11915742,
        0.0088214455,
        0.03348277,
        -0.0357388,
        0.0005325822,
        0.025311925,
        0.08477533,
        0.028816674,
        -0.043864574
      ]
    },
    {
      "id": "worldsfair-speaker-383",
      "conference": "worldsfair",
      "name": "Theo Browne",
      "role": "",
      "company": "t3.gg",
      "twitter": "https://x.com/theo",
      "talks": [
        {
          "title": "Closing Keynote — Theo Browne",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "4:30pm-4:50pm",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.017718947,
        0.028313149,
        0.037602197,
        -0.009252461,
        -0.006150602,
        -0.01686088,
        -0.04397267,
        0.011369888,
        0.0467937,
        -0.18832244,
        -0.03910363,
        -0.043292858,
        0.028360557,
        -0.007952245,
        -0.019632498,
        0.03945067,
        0.0679841,
        -0.07168285,
        -0.058202777,
        -0.11245016,
        -0.05838549,
        -0.053237274,
        -0.027010974,
        0.048820183,
        -0.077906504,
        0.09995036,
        -0.017433424,
        0.0266833,
        -0.13184106,
        0.5029912,
        -0.03793548,
        -0.008044299,
        0.031207401,
        0.0008379212,
        0.04546551,
        -0.015699716,
        0.04210165,
        0.10151315,
        0.042649496,
        -0.009002258,
        -0.016108684,
        0.009325011,
        -0.018237712,
        0.049134023,
        0.05148808,
        0.04871967,
        -0.08089847,
        -0.08529372,
        -0.0061319266,
        0.07188265,
        -0.07982938,
        0.031551618,
        -0.020370038,
        -0.013093656,
        0.06565538,
        -0.025362073,
        0.011642699,
        -0.053431604,
        0.075709425,
        -0.0055234525,
        -0.1344135,
        -0.07486091,
        -0.019079862,
        -0.030148245,
        0.0061037843,
        -0.017052222,
        0.07954959,
        -0.024144862,
        0.044411156,
        -0.06618146,
        0.0127392495,
        -0.0031589887,
        -0.006178,
        0.015760012,
        -0.07552343,
        -0.02082343,
        -0.020314772,
        0.06408171,
        0.03169736,
        0.019847477,
        0.04341178,
        -0.008219089,
        -0.05232519,
        -0.07901558,
        -0.019499946,
        -0.045532763,
        -0.02717503,
        0.03232977,
        0.037270255,
        -0.004584875,
        -0.013173839,
        0.010163988,
        -0.07361614,
        0.012309156,
        -0.05691529,
        0.0013509783,
        -0.04329174,
        -0.044497333,
        -0.032051466,
        -0.059336264,
        -0.042049233,
        -0.039521463,
        0.02325103,
        -0.03233557,
        -0.05806564,
        -0.054602627,
        -0.02315355,
        0.010908747,
        0.023629503,
        0.008107009,
        -0.050299004,
        -0.642216,
        0.022202572,
        0.031544656,
        -0.008558602,
        -0.15019579,
        -0.03667486,
        -0.006673757,
        -0.01594654,
        0.05499791,
        0.013844538,
        0.07509593,
        -0.012766447,
        0.0828628,
        -0.009841597,
        0.0615643,
        0.04174484,
        -0.032132253
      ]
    },
    {
      "id": "worldsfair-speaker-384",
      "conference": "worldsfair",
      "name": "Garry Tan",
      "role": "",
      "company": "",
      "twitter": "",
      "talks": [
        {
          "title": "Closing Keynote: Garry Tan",
          "description": "",
          "day": "Day 4 — Session Day 3",
          "time": "4:50pm-5:10pm",
          "type": "keynote"
        }
      ],
      "embedding": [
        0.038784515,
        0.046094745,
        -0.012040407,
        0.0012964676,
        0.059856426,
        -0.00034606445,
        -0.1079885,
        -0.006408061,
        0.06963129,
        -0.17010236,
        0.01719181,
        -0.007946309,
        -0.051549092,
        -0.053524535,
        -0.038058132,
        0.023402147,
        0.001094639,
        -0.083014384,
        -0.07009851,
        -0.1345142,
        0.030966418,
        -0.053964145,
        -0.031237597,
        0.012576304,
        -0.047806744,
        0.15482546,
        -0.025972558,
        0.0557601,
        -0.11817411,
        0.44526216,
        0.06532096,
        -0.028421573,
        -0.020265803,
        -0.038794868,
        -0.032716207,
        -0.044730045,
        0.019390743,
        0.08275492,
        0.048923895,
        0.071380086,
        0.015279655,
        -0.00084883417,
        -0.023992913,
        0.048879363,
        -0.07239093,
        -0.011446917,
        -0.17144799,
        -0.10502828,
        0.01201766,
        0.0068029123,
        -0.063031085,
        0.04688716,
        -0.047109548,
        -0.04621363,
        -0.012967232,
        -0.010235813,
        -0.07851683,
        -0.017808883,
        0.109136455,
        -0.043533586,
        -0.15012407,
        0.024168473,
        0.019087976,
        -0.015435092,
        -0.060520943,
        -0.0059167156,
        0.07223798,
        0.006600051,
        0.019143926,
        -0.042498037,
        0.012009859,
        -0.0013976656,
        -0.0012882377,
        0.0060423426,
        -0.122795865,
        0.022938147,
        -0.00448956,
        0.008216508,
        -0.01991319,
        -0.009677272,
        -0.0079990765,
        -0.017887553,
        -0.09915634,
        -0.030449834,
        -0.034631122,
        -0.07432947,
        -0.0065091597,
        0.053243984,
        -0.0052873567,
        0.1265227,
        -0.005278657,
        0.00096128363,
        -0.07343555,
        -0.035671625,
        -0.069161125,
        0.07081441,
        -0.1039764,
        0.060565114,
        -0.09841699,
        0.025551757,
        0.010272749,
        0.016648296,
        -0.008569178,
        -0.026693733,
        -0.07664599,
        -0.020651335,
        -0.0341173,
        -0.009470276,
        0.030847216,
        0.012740723,
        -0.0006267376,
        -0.60743666,
        0.011331024,
        0.009613443,
        -0.012804384,
        -0.13459583,
        -0.06991338,
        -0.026480813,
        0.051182408,
        0.07546242,
        0.026787682,
        0.05464826,
        0.032727066,
        0.040510263,
        -0.037473578,
        0.09119882,
        0.0028572122,
        0.01950342
      ]
    }
  ]
}