Nat Gurlain
Software engineer with a deep interest in artificial intelligence, machine learning, and building tools that push the boundary of what autonomous systems can do.
Passionate about reinforcement learning, deep learning, and crafting open-source software that solves real problems.
Projects
Multi-chain staking explorer with real-time validator analytics, reward calculators, and portfolio tracking for Proof-of-Stake networks.
Workspace-first AI guardian with persistent memory, proactive scheduling, desktop awareness, tool use, reusable workflows, and a live browser cockpit for operator control.
Codex-first control plane for supervising autonomous work, with governed execution loops, explicit review and recovery, project memory, and observable runs, logs, traces, and incidents.
Lightweight macOS website blocker that automatically restricts distracting sites during configurable work hours using system-level hosts file manipulation and cron scheduling.
DQN-to-Rainbow for Atari: incremental deep reinforcement learning implementations with experiment tracking via MLFlow.
Interests
Deeply curious about artificial intelligence, reinforcement learning, and deep learning — from training autonomous trading agents to exploring the frontiers of language models. Also drawn to quantitative finance, blockchain technology, systems programming, and the art of competitive programming. Writes about the intersections of AI, mathematics, and financial markets on Substack.