Innodata Hackathon: Build at the Frontier of AI
Join Innodata for a full-day, in-person hackathon in New York City on Saturday, November 14.
At Innodata, we work on some of the hardest problems at the frontier of AI, from creating and evaluating advanced AI systems to building the data, benchmarks, and engineering workflows that make them better. We're bringing together a select group of engineers and AI builders to take on real technical challenges in that space.
You'll join a team of 4-5 to design, build, test, and demo a working prototype, using organizer-provided model APIs and AI coding assistants. We care about strong engineering judgment, creative problem solving, and evidence that your approach works beyond a single example. A focused, reliable prototype wins over a polished product.
Tracks
Coding Task Factory: How can we automatically create realistic software-development tasks that are clear, solvable, and reliably testable? Build a reusable pipeline that generates feature-development tasks for evaluating AI coding systems, complete with requirements, reference implementations, and tests.
Visual Task Factory: Which visual capabilities matter most in real-world applications, and how can we test whether frontier models have them? Build a pipeline that generates synthetic images, questions, and verifiable answers, then use it to probe frontier multimodal models.
Prizes
1st place: $5,000
2nd place: $2,500
3rd place: $1,250
Schedule
Thursday, November 12: Team pairings announced
Friday, November 13, 6:00–8:00 PM: Welcome party and happy hour (optional; location shared upon acceptance)
Saturday, November 14, 8:30 AM: Registration opens
Saturday, November 14, 9:00 AM–5:00 PM: Hackathon, demos, and judging
What to bring
A laptop with a working development environment, and familiarity with API calls, dependency management, and automated testing. Python or another suitable language is fine, and no specialized hardware is needed.
Who should apply
Curious, hands-on engineers and AI builders who like tackling unfamiliar problems, testing what works, and figuring out why. Experience with AI evaluation, synthetic data, or code-generation workflows is helpful but not required.
Come ready to build, break things, debug, and rethink what AI systems can do.
Please note: space is limited and participation is by application. Accepted participants will be asked to sign a participant agreement before the event. Travel expenses are the participant's responsibility.