About Me

I build at the intersection of AI engineering and applied cryptography — LLM tooling that runs in production, and more recently FHE and zero-knowledge proofs, where I care about what they can actually do rather than promise.

I care about shipped software. Demos don't count until real users are on them, and most of what I know came from running systems in production — the successes, and more often the failures.

This site is where I write it down: field notes on what I'm building and where I got it wrong, so neither of us has to make the same mistake twice.

The work so far includes PixelsAI, an AI image-generation service carrying production traffic, and contracts deployed to mainnet. Current focus: fully homomorphic encryption, LLM agents that do real work, and the tooling in between.

If you're building something in this space, say hi on X — I'm always up for comparing notes on production systems, or finding an excuse to build something together.