Nat Gurlain
Software engineer building AI products and open-source tools.
I work mostly on AI agents, reinforcement learning, and tools that help people learn and think, with side interests in quantitative finance and blockchain. I write about AI, mathematics, and markets on Substack.
Products
AI flashcards made from your own material, with mnemonics for the tricky parts and spaced repetition to make it stick.
Books recommended by great builders, investors, and thinkers, with every recommendation linked to its source.
Check your NEAR stake, estimated rewards, validator health, and fees without connecting a wallet.
Compares estimates of the world's AI compute with the compute needed to emulate a human brain, using Epoch AI data.
Open source
An AI assistant with long-term memory that plans ahead, uses your tools and browser, and keeps you in control of what it does.
A control plane for supervising Codex coding agents, with governed runs, review and recovery, project memory, and full logs and traces.
A small macOS tool that blocks distracting websites during your work hours.
Deep reinforcement learning for Atari, built up step by step from DQN to Rainbow, with MLflow experiment tracking.