Anthropic · 2026-05-19 · major
Andrej Karpathy Joins Anthropic — OpenAI Co-Founder Lands on the Pre-Training Team to Build a New Group That Uses Claude to Accelerate Pre-Training Research
Andrej Karpathy announced he has joined Anthropic and is starting this week on the pre-training team under Nick Joseph, where he will lead a new group focused on using Claude to accelerate pre-training research itself.
Karpathy moves from Eureka Labs to Anthropic's pre-training team to lead a Claude-assisted pre-training research group.
What is it?
Andrej Karpathy — a co-founder of OpenAI, former director of AI at Tesla, and most recently the founder of education startup Eureka Labs — has joined Anthropic. He posted on X that the next few years at the frontier of LLMs will be "especially formative" and that he is excited to "get back to R&D." Anthropic confirmed he starts this week on the pre-training team under team lead Nick Joseph.
How does it work?
Pre-training is the large-scale training phase that gives a model its core knowledge and capabilities. Karpathy's brief is to build a new team that uses Claude itself to accelerate pre-training research — running experiments, generating and triaging training data, and pushing more of the research loop through agents instead of human researchers. The move is Anthropic's bet that research productivity (not just raw compute) is now the rate-limiting step at the frontier.
Why does it matter?
Karpathy is one of the best-known LLM communicators alive (nanoGPT, Zero to Hero, llm.c) and a marquee hire from OpenAI's founding team. The signal: leading researchers are increasingly betting that AI-for-AI-research compounds faster than pure scaling. The hire also lands while Anthropic is closing a reported $30B funding round and just acquired Stainless, the company behind every official Claude SDK.
Who is it for?
AI researchers, frontier lab watchers, talent-market observers
Try it
Read Karpathy's X post at https://x.com/karpathy