Long Horizon Agents Hackathon

Long-horizon agents still break down over time: they accumulate observations, actions, and stale context until the growing history slows them down, costs more, and makes their own state less reliable.

Fixing that means rethinking the architecture — explicit mutable state instead of ever-growing histories, agents that edit their own working context, and a clear split between what needs to persist and what can be discarded. This hackathon is about building that architecture.

Build systems that help agents preserve what matters and discard what doesn't, staying reliable as tasks stretch from minutes to hours, days, or longer.

We're backing you with some of the best tools

AWS: The models and cloud infrastructure to help you actually ship

Airbyte: The Context Layer for AI Agents

Liquid AI: Efficient general-purpose AI at every scale.

Nimble: Web data your agents can trust

Speakers

Viviana Márquez - Developer Relations @ Liquid AI

Tianshu Yu- Member of Technical Staff, ML Engineer @ Liquid AI

Yaniv Markovski - Head of Ecosystem Engineering @ Nimble

Loading...

Judges

Viviana Márquez - Developer Relations @ Liquid AI

Tianshu Yu - Member of Technical Staff, ML Engineer @ Liquid AI

Yaniv Markovski - Head of Ecosystem Engineering @ Nimble

Mogana Kumaran S. - Senior Staff Data Engineer @ Gap INC

Amit Panda - Staff Software Engineer @ LinkedIn

Loading...

Agenda

• 9:30 AM — Doors Open + Opening Remarks

• 11:00 AM — Kickoff & Hack

• 1:30 PM — Lunch

• 4:30 PM — Project Submission

• 5:00 PM — Finalist Demos + Judging

• 7:00 PM — Awards + Closing

Thank you to our ​Partners!

Note: This event is fully in-person in San Francisco. Capacity is strictly limited. All attendees must apply and be approved prior to the event. Unapproved or pending applicants will not be admitted at the door.

Maximum team size is four. 18+ only

See what’s possible.

Follow us on X · LinkedIn · Newsletter