Who is Andrej Karpathy, an OpenAI pioneer who just swapped sides to Anthropic?

Andrej Karpathy, one of the founding members of OpenAI and a former artificial intelligence executive at Tesla, has joined rival AI firm Anthropic, marking one of the most closely watched talent moves in the global artificial intelligence industry.

Karpathy confirmed the move in a post on X, saying he was “very excited” to return to research and development as competition intensifies among companies building large language models (LLMs). According to the X post, he has joined Anthropic’s pretraining team, which is responsible for the large-scale model training that powers Claude, the company’s flagship AI assistant. 

The move comes at a time when AI companies are aggressively competing for engineers and researchers capable of building next-generation foundation models. Anthropic, founded in 2021 by former OpenAI executives including Dario and Daniela Amodei, has emerged as one of OpenAI’s biggest rivals in enterprise AI and generative AI systems.

From OpenAI and Tesla to the centre of the AI race

Karpathy is widely recognised in the artificial intelligence community for his work in deep learning, neural networks and autonomous driving systems. He studied under Stanford AI scientist Fei-Fei Li and was among OpenAI’s earliest researchers before joining Tesla, where he led computer vision and AI efforts tied to the company’s Autopilot technology, which is one of Tesla's most important AI divisions.

Karpathy is also the one who coined the term “vibe coding” in 2025. Vibe coding is a software development practice where you build applications by describing your ideas in plain, natural language to an AI assistant.

After leaving Tesla in 2022, Karpathy became increasingly known for his educational work on AI, publishing technical explainers and tutorials that gained significant traction among developers and researchers. In 2024, he launched Eureka Labs, an AI-focused education platform aimed at integrating AI tools into learning systems. 

Anthropic said Karpathy would work under Nick Joseph, the company’s head of pretraining. The division focuses on large-scale training runs that determine the reasoning capabilities and knowledge base of Claude models. Reuters reported that Karpathy started work this week. 

His appointment also highlights a broader reshaping of leadership across the AI industry. Several high-profile figures linked to OpenAI have departed in recent years, including former chief scientist Ilya Sutskever, former chief technology officer Mira Murati and co-founder John Schulman, who also joined Anthropic in 2024.