I build at the edges of crypto and AI — the parts that are still figuring out what they are.
Most weekends I'm shipping something at a hackathon. Most weekdays I'm writing the spec for the thing I'll ship next weekend.
Currently: building something I'll talk about when it works.
I believe the interesting problems live between specialties, so that's where I work.
▸ chat-first interfaces for on-chain actions
▸ tokenized real-world assets and the compliance rails around them
▸ AI agent coordination (the boring infra parts, not the hype parts)
▸ cross-chain settlements, MPC, intents, atomic swaps, the whole bestiary
I write the spec first. A long one. The kind where invariants, edge cases, and threat models are pinned down before a single line of code exists. Then I hand it off to a team of Claude Code agents and clean up what they produce until it's something I'd sign my name to.
I'd rather be wrong fast than right slow. Push back on me — I'll respect it more than agreement.
research → spec → build → ship → post-mortem → repeat
Deadlines I invent for myself are the only ones that work. Hackathons are the deadlines.





