He helped build OpenAI, he ran Tesla’s self-driving program, and now Andrej Karpathy is heading to Anthropic. Karpathy confirmed the news on Tuesday in a post on X.
“I’ve joined Anthropic,” he wrote. “I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D.”
Job Role
Karpathy is joining Anthropic’s pre-training team. That’s the group responsible for the massive training runs that shape Claude’s core knowledge and abilities.
Think of pre-training like the foundation of a house. Everything else, the safety features, the fine-tuning, the user experience, gets built on top of it. Get the foundation wrong, and everything suffers.
He’ll work under Nick Joseph, who leads the pre-training team. But Karpathy won’t just be another member of the crew.
Anthropic says he’ll build a brand-new team focused on using Claude to speed up pre-training research itself.
Using AI to make AI better at its own training process is the kind of recursive thinking that only a handful of people in the world are qualified to lead.
And Karpathy is one of them.
Anrej Karpathy
Anrej Karpathy is one of the most respected AI researchers alive. His ability to connect deep theoretical knowledge with real, large-scale training practice is genuinely rare.
Pre-training is also one of the most expensive parts of building a frontier AI model. It burns enormous amounts of compute.
If Karpathy’s team can find smarter, AI-assisted ways to do that research, Anthropic could close the gap with bigger-budget rivals like OpenAI and Google, without just throwing more hardware at the problem.
That’s a clever strategy, and hiring Karpathy to lead it sends a clear message about where Anthropic thinks the real edge will come from.
His Career
Karpathy co-founded OpenAI back in 2015. While there, he worked on deep learning and computer vision.
He left in 2017 to join Tesla, where he eventually led the Full Self-Driving and Autopilot programs, two of the most ambitious real-world AI efforts outside of a research lab.
He left Tesla in 2022, then he returned to OpenAI for about a year before stepping away again in 2024.
That second exit was to launch Eureka Labs, a startup focused on bringing AI assistants into education. The vision was to create AI tutors that could genuinely help people learn.
But Eureka Labs has been quiet since its launch. And now Karpathy is at Anthropic.
Eureka Labs
Karpathy hasn’t said the startup is over. But he also hasn’t shared much about its progress. His move to a full-time role at Anthropic raises obvious questions about where his focus will be.
He did address it, briefly. “I remain deeply passionate about education and plan to resume my work on it in time,” he said.
Even with a big new role, Karpathy has built something that keeps running without him.
His online course, Neural Networks: Zero to Hero, has helped thousands of people learn to build neural networks from scratch using real code.
His YouTube channel is similar, a mix of lectures and walkthroughs on LLMs and AI that are dense, clear, and genuinely useful. It’s the kind of teaching that doesn’t talk down to you.
That content will keep reaching people whether or not he posts new material anytime soon.
