Co-founder of deterministic labs. We build machine intelligence whose answers are made to be checked, not believed. First product ships soon.
I write at circuit about the mathematics inside neural networks, starting with a six-part series on superposition. The lab writes too: firm ground.
Open source:
- linearly: a free, visual-first linear algebra course for people building AI, following MIT 18.06; I wrote and taught the lectures live in 2022 and 2023, before the LLM era
- deepsense: drug discovery with graph and convolutional networks
- wabbitc: a compiler with two backends, LLVM and WebAssembly encoded from the core spec
- aioraft: the Raft consensus algorithm on asyncio, implemented from the paper
- aiodiameter: a pure-Python Diameter protocol stack (RFC 6733), written against a live telecom charging system

