Andrej Karpathy is an 📝Artificial Intelligence (AI) 🏷️#researcher and 🏷️#educator widely regarded as one of the most influential public voices in 📝Machine Learning (ML). He was a founding team member of 📝OpenAI (2015), Director of AI at 📝Tesla Motors from 2017 to 2022 where he led Autopilot and the Dojo supercomputer program, and is founder of 📝Eureka Labs (2024), an AI-native education company. He earned his 🏷️#phd in computer vision and deep learning at 📝Stanford University under Fei-Fei Li, designed Stanford's first deep learning course, and has built an audience of over a million followers on X around his open-source work and long-form teaching.
His independent practice has produced a string of artifacts now treated as canon in the field: the Neural Networks: Zero to Hero video series, the micrograd and nanoGPT reference repos, and the autoresearch autonomous-training framework. He coined 📝vibe coding in February 2025 and the more disciplined 📝agentic engineering framing in February 2026, naming shifts the industry adopted within weeks. His April 2026 post on LLM-maintained knowledge bases (see 📝Karpathy's Flag in the Ground — LLM Knowledge Bases) is the public flag most often pointed to when explaining what MythOS has been building since 2017.
I have never met Karpathy, but his public arc is the closest external validation of the MythOS thesis I have found. When he names a pattern, the field rearranges around it within weeks — which is why he anchors 📝The Karpathy Weathervane Pattern, and why 📝Why MythOS Is What Karpathy Described is the cleanest articulation of the overlap with my own work.