Andrej Karpathy Joins Anthropic to Lead LLM Team

Andrej Karpathy announced on Tuesday that he has joined Anthropic and will focus on large language model training and systems that enable autonomous agents. In a post he wrote, “I’ve joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative… I remain deeply passionate about education and plan to resume my work on it in time.” The company has not disclosed his team or specific projects.

Karpathy is a well-known AI researcher who helped found OpenAI in 2015. He left OpenAI in 2017 to work at Tesla, returned to OpenAI in 2023 for about a year, and departed again in February 2024. After that exit he worked independently, starting an education venture called Eureka Labs and producing a popular video series on neural networks and language models.

In February 2025 Karpathy introduced the term “vibe coding” to describe a form of AI-assisted programming in which developers accept model-generated changes without reading diffs. He later described a related approach, which he calls agentic engineering: human users set specifications and oversee work while autonomous agents carry out multi-step tasks.

Anthropic was founded in 2021 by former OpenAI researchers and develops the Claude/Opus family of models. The company released Opus 4.7 in April with upgrades for long-form reasoning and vision. OpenAI released GPT-5.5 days later, positioned for autonomous, multi-step work. Both firms have made a number of senior hires in recent quarters.

Karpathy’s past research includes neural network architecture, computer vision and synthetic data, areas relevant to large-model training and agent design. He has said he plans to continue his public education work while at Anthropic. Neither he nor Anthropic has provided a timetable or detailed responsibilities for his role.