I build software.
I want to do research.

I started by shipping small products: browser extensions, developer tools, and data systems. That taught me how to debug, scope work, and finish things.

I also worked in finance. I met an extraordinary concentration of maths and physics talent there, but I kept thinking that using so much human ability to find smaller market edges was a waste. It made me certain that I want to work in technology and aim my own effort at helping people more directly.

I want to live a magnificent life. I do not mean a comfortable one. I mean attempting difficult work, making something genuinely useful, and doing some good while I am here.

Neural decoding feels like magic to me. I would love to think an instruction and have it happen: tell an AI agent to play a song through a speaker, communicate without speaking, or control a machine without moving. That direct path from thought to action is the first thing I want to work on.

World models come second. Language models are powerful, but I think their ceiling is lower. A system that learns how an environment works, predicts what its actions will cause, and plans within it could become the most useful application of AI.

I have not published ML research or done a PhD. The gaps are obvious: maths, neuroscience, and experimental design. I am working on them.