Tools for the AI-assisted future. Systems that compose. Infrastructure that doesn't surprise you.

I care about correctness, reproducibility, and building things that last.

Started in embedded systems — where every byte matters and bugs crash hardware, not just processes.

Moved through iOS — mobile at scale, real users, real constraints.

Went deep on functional programming — type systems as documentation, composition as architecture, reasoning about code without running it.

Now building AI/ML infrastructure — because the same principles that make distributed systems reliable apply to making LLMs useful.

The common thread: systems where correctness matters.

I contribute to tools that help developers be more effective.

• infiniloom — High-performance repository context generator for LLMs. Transform codebases into optimized formats for Claude, GPT-4/5, Gemini.

• More projects at Topos Labs — AI research & development.

• Correctness over convenience — The easy path is rarely the right path
• Composition over inheritance — Build small things that combine well
• Explicit over implicit — Magic is great until it isn't
• Delete more than you add — Simplicity is a feature

"The purpose of abstraction is not to be vague, but to create a new semantic level in which one can be absolutely precise." — Edsger W. Dijkstra