Andrej Karpathy Joins Anthropic to Lead AI Pretraining Research

Andrej Karpathy, a Slovak-Canadian artificial intelligence researcher and co-founder of OpenAI, announced on May 19, 2026, that he has joined the AI lab Anthropic. Karpathy, 39, previously served as the head of Tesla’s AI division and has been a prominent figure in the development of large language models (LLMs) and computer vision.

The move places Karpathy in a strategic role within Anthropic’s Pretraining team. According to Nicholas Joseph, Anthropic’s Head of Pretraining and a former OpenAI employee, Karpathy will lead a new team dedicated to using the Claude model to accelerate the process of pretraining research.

Excited to welcome Andrej to the Pretraining team! He’ll be building a team focused on using Claude to accelerate pretraining research itself. I can’t think of anyone better suited to do it — looking forward to what we build together!

Nicholas Joseph

Karpathy confirmed the transition via a post on the social network X on May 19, 2026, stating that he believes the next several years at the frontier of LLMs will be especially formative and expressing his enthusiasm to return to research and development.

Personal update: I’ve joined Anthropic. 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. I remain deeply passionate about education and plan to resume my work on it in time.

Andrej Karpathy

The announcement coincided with the start of Google’s annual I/O developer conference in Mountain View, California, a period typically marked by significant product releases and technical announcements from the AI industry.

A career across research, deployment, and education

Karpathy’s professional trajectory spans the primary pillars of the current AI boom: academic research, large-scale corporate deployment, and public education. He was one of the original 11 co-founders of OpenAI, listed in the organization’s launch announcement in December 2015.

From 2017 to 2022, Karpathy served as the Director of AI at Tesla, where he led the computer vision team for the Autopilot system. In this role, he managed neural network training, the deployment of models on Tesla’s custom inference chips, and in-house data labeling efforts.

He returned to OpenAI from 2023 to 2024, focusing on the creation of a team centered on synthetic data generation and midtraining. This specific experience is closely aligned with his new responsibilities at Anthropic regarding pretraining research.

Karpathy’s technical foundation was established at Stanford University, where he earned a PhD under the supervision of Fei-Fei Li. His doctoral work focused on the intersection of natural language processing and computer vision. His academic credentials also include a Master of Science from the University of British Columbia and a Bachelor of Science in computer science and physics from the University of Toronto.

Throughout his career, he has also held internships at DeepMind, Google Research, and Google Brain, and he helped develop Stanford’s first deep learning course, CS231n.

Open source contributions and AI education

Since leaving OpenAI in 2024, Karpathy has operated as a free agent, focusing heavily on AI education and open source research. In July 2024, he founded Eureka Labs, an AI-native school that released LLM101n, an undergraduate-level course designed to teach students how to train their own AI systems.

During this period, he also developed several open source tools and standards, including:

• autoresearch: An LLM-driven automated researcher capable of running multiple experiments and hypotheses simultaneously.
• LLM Knowledge Base: An autonomous system designed to store memory and context for AI agents in an expandable library.

Karpathy’s move to Anthropic raises questions regarding the future of these open source projects. While Anthropic has supported open standards through the launch of the Model Context Protocol (MCP), the company primarily develops proprietary AI models and tools, such as Claude and Claude Code.

Regarding his educational initiatives, Karpathy indicated in his May 19, 2026, announcement that he plans to resume his work in education in time, suggesting that his efforts with Eureka Labs may be paused while he integrates into his new role at Anthropic.