Overview
AI researcher and educator. Co-founder of OpenAI; former Director of AI at Tesla (led Autopilot). Known for influential technical writing, lectures (Stanford CS231n), and open-source contributions that make deep learning concepts accessible. In April 2026, published a GitHub Gist titled "LLM Wiki" — a conceptual design pattern for LLM-maintained personal knowledge bases — that sparked wide discussion in the AI community.
Key Contributions / Features
- LLM Wiki (2026): Published as a GitHub Gist, not an app or library — described as an "idea file" meant to be copy-pasted into an agentic LLM tool like Claude Code or OpenAI Codex. The pattern proposes a three-layer architecture (immutable raw sources, LLM-maintained markdown wiki, schema/CLAUDE.md operating manual) with three operations (ingest, query, lint). Core claim: inserting a compounding pre-compilation layer between raw documents and queries solves both the statelessness of RAG and the maintenance collapse of human-curated wikis. Mental model: "Obsidian is the IDE, the LLM is the programmer, the wiki is the codebase." See LLM Wiki.
Related
Sources