OC AI Builders Lightning Hackathon: Decisions + LLMs
What can you build with AI in 50 minutes? Come find out, then show the room what you made. Part of OC Tech Week.
The format is simple: you get 50 minutes to build something with AI, then up to 5 minutes to demo it. Come with an idea or start from scratch. Build solo or form a small team when you arrive. Rough, weird, incomplete experiments are welcome. The point is to make something real, show it, and trade notes.
SUGGESTED THEME: DECISION MODELS + LLMs
Decision models recently burst on the scene: models like Jev from Typesafe AI that answer a focused question — should this action happen? is this request normal? — with a calibrated judgment instead of free-form text. Try pairing one with an LLM: let the LLM generate, and let the decision model (plus your own code) make the call. Projects using just regular LLMs are totally welcome too — the theme is a prompt, not a rule.
FORMAT
• 50 minutes to build
• 5 minutes to demo per team
• Solo builders and small teams are welcome
• A small working slice beats a pitch deck
AGENDA
• 5:00–5:30 PM — Arrival, food, drinks, and networking
• 5:30–5:40 PM — Welcome, theme, format, and team-up
• 5:40–6:30 PM — 50-minute build sprint
• 6:30–7:30 PM — Five-minute demos
• 7:30–8:00 PM — Food, drinks, networking, and follow-ups
WHAT CAN YOU BUILD?
Agent guardrails and policy checks, escalation or triage bots, approval workflows, classifiers and routers, evals, LLM-as-judge pipelines, developer tools, or something we haven't thought of yet.
WHO SHOULD ATTEND
• Engineers and founders building with AI
• Product and operations people turning AI experiments into real workflows
• Curious builders who want candid technical discussion, not a sales pitch
WHAT TO BRING
Bring a laptop, charger, an idea if you have one, and a willingness to ship something small. Food and drinks will be provided. Open models through Nebius Token Factory are one option for your build. [PLACEHOLDER: confirm whether Nebius will provide credits]
This event is presented by Vivek Haldar, a proud Nebius Fellow, with support from Nebius. https://nebius.com/fellows
Explore the Nebius AI Builder Program: https://dev.nebius.com/builders