Travis Law
I build software and hardware for the fun of finding out whether it'll work. Lately that means pointing all of it at a treasure hunt that's gone unsolved since 1982. LawLab is the notebook.
01About
I'm Travis, a builder based in Indiana, USA. My projects tend to live where Python meets Raspberry Pi-class hardware and, more recently, AI that runs on my own machines. I like local-first systems, structured data over vibes, and builds with a physical payoff: altitude, candy, or treasure.
Past experiments have included CO₂-powered candy artillery for trick-or-treaters, a flight recorder for model rockets, and a touchscreen panel that runs my desk. The current obsession is bigger: a full retrieval-and-agents research system aimed at one of America's longest-running treasure hunts.
When the soldering iron is cold, I'm usually watching the Blues or the Cardinals.
02Featured project
Active · RAG + Agents
Hunting The Secret
In 1982, Byron Preiss buried twelve ceramic casques across North America and hid the directions inside twelve paintings and twelve verses. Three have been found. This project loads everything publicly known about the other nine into a local RAG system: CLIP-indexed paintings, verse hypotheses, a tiered-provenance corpus, and research agents that stage theories for human review. Recent output has checked out on real ground, with more field trips planned.
- Python
- ChromaDB
- CLIP
- local Qwen3-32B
- Claude Opus & Fable
- OpenAI cross-checks
- SQLite
The rest of the bench, candy cannon included: all projects →