An in-person flagship at 30 Hudson Yards, New York — convening a global audience of world-class data and AI engineers. Four days where the field's most capable builders come together to design, train, and ship production-grade intelligence. No tutorials. No filler. Only builders operating at the frontier.
This is not an open hackathon. It is a curated gathering of world-class data and AI engineers — industry experts, international professionals, founding engineers, research leads, and CTOs — brought together in New York to push what AI products can do across four intense days.
Every participant is reviewed before being admitted. We protect the bar relentlessly, because the quality of the room is the entire point. You will be building beside people whose work you already respect.
If you see yourself here, you should apply.
Teams choose a track on day one and ship a working system by the close. Cross-track ambition is welcome — the only rule is that it must run.
Multi-step, tool-using agents that plan, act, and self-correct on real tasks. Reliability and guardrails over demos.
Next-generation RAG, hybrid search, and knowledge graphs that stay grounded, fresh, and verifiable at scale.
Streaming, transformation, and lineage systems that make AI trustworthy — the unglamorous layer that decides everything.
Low-latency, cost-aware model serving: quantization, batching, caching, and routing across heterogeneous hardware.
Rigorous eval harnesses, red-teaming, and live observability for AI products you can actually trust in production.
Human-facing intelligence: interfaces, copilots, and experiences that turn raw model capability into real product value.
In person at 30 Hudson Yards, New York, NY 10001. A continuous build window from open to demo. We keep the schedule deliberately light — the work sets the pace, not a timetable.
Doors open at Hudson Yards. Teams assemble, tracks are chosen, problem framing and architecture sprints. The build begins.
Heads-down engineering. Mentor and judge office hours throughout. Systems take shape and get stress-tested.
Integration, evaluation, and hardening. Teams push their systems toward production reliability.
Final commits, live demos, and a rigorous judging round. Winners announced and celebrated.
Our judging panel carries as much weight as our participants. Every judge passes a rigorous, multi-stage interview process before they are allowed to evaluate a single project.
Admission is competitive and reviewed individually. Tell us who you are and we'll be in touch.
If you have the depth to evaluate world-class AI and data work, we want to meet you. Selected applicants enter our vetting and interview process.
To be considered as a judge, email stephen@aiprodatahack.com with why you want this judging spot and your bio.
Selected judges are then contacted to begin the interview and vetting process.
The inaugural edition set the bar. Here's what the community built before we brought the flagship to New York.
July 8–10, 2026 · Virtual · Global
Our first edition ran fully online, proving the format: a tightly vetted room of world-class engineers shipping production-grade AI in 72 hours. It seeded the network that now convenes in person at 30 Hudson Yards, New York.