Summary
A two‑day, dev‑focused AI hackathon in Seattle oriented around building with agents, skills, and modern AI developer tooling—team up, ship something, and learn the implementation patterns in the process.
Why it matters
Hackathons compress the learning curve: you hit the real problems fast (tool reliability, prompt drift, evals, and latency). That makes them a good forcing function for practitioner-level techniques instead of slide-deck AI.
Tokenmaxxing angle
If you want to use a lot of AI, you need constraints early: stop conditions, caching, retrieval boundaries, and basic eval loops. Build nights are the fastest way to develop those instincts before you scale token spend in production.
Source takeaway
Expect hands-on time with agents/tooling plus concrete heuristics for keeping agent loops reliable and cost-aware.