Andrej Karpathy
☑️
Pre-training Researcher, Anthropic
at
OpenAI
Andrej Karpathy is a Slovak-Canadian AI researcher who joined Anthropic's pre-training team in May 2026, where he leads research on using Claude to accelerate pre-training itself. He previously founded the AI-education startup Eureka Labs after two stints at OpenAI and five years as Tesla's Director of AI.
Karpathy is one of OpenAI's co-founders (2015) and led the neural networks behind Tesla Autopilot and Full Self-Driving from 2017 to 2022. At Stanford he created CS231n, the university's first deep learning course, and his 'Neural Networks: Zero to Hero' video series and nanoGPT codebase have taught a generation of engineers to build language models from scratch. He coined the term 'vibe coding' in 2025.
His move to Anthropic, betting his research career on the claim that pre-training is not done, was one of the most-watched AI hires of 2026.
Karpathy is one of OpenAI's co-founders (2015) and led the neural networks behind Tesla Autopilot and Full Self-Driving from 2017 to 2022. At Stanford he created CS231n, the university's first deep learning course, and his 'Neural Networks: Zero to Hero' video series and nanoGPT codebase have taught a generation of engineers to build language models from scratch. He coined the term 'vibe coding' in 2025.
His move to Anthropic, betting his research career on the claim that pre-training is not done, was one of the most-watched AI hires of 2026.
Background & Achievements
Education
PhD, Computer Science, Stanford University; MSc, University of British Columbia; BSc, Computer Science and Physics, University of Toronto
Key Achievements
- 🏆 Co-founded OpenAI (2015) as a founding research scientist
- 🏆 Director of AI at Tesla (2017-2022); led the vision neural networks for Autopilot and Full Self-Driving
- 🏆 Founded Eureka Labs (2024), an AI-native education startup; created Stanford's CS231n course and the 'Zero to Hero' neural-net series
- 🏆 MIT Technology Review Innovators Under 35 (2020); TIME100 AI (2024)
Expertise
large language models
model pre-training
computer vision
ai education
autonomous driving