Executive Spotlight: Andrej Karpathy, AI Researcher on Pretraining at Anthropic (2027)
Andrej Karpathy is an AI researcher on Anthropic's pretraining team. He announced on X on May 19, 2026 that he had joined Anthropic, and an Anthropic spokesperson told VentureBeat he will start a team focused on using Claude to accelerate pretraining research. Nicholas Joseph, Anthropic's head of pretraining, welcomed him to the team on X.
He is widely known for three earlier chapters: a founding member of OpenAI in 2015, Director of AI at Tesla from 2017 to 2022, and founder of Eureka Labs, an AI-native school announced in July 2024. He also built and taught Stanford's CS231n deep learning course while a PhD student.
The role was checked on October 31, 2026. His own site karpathy.ai still lists his work as education videos and does not yet mention Anthropic, so the current role rests on his X post as reported by VentureBeat and on other press coverage. This page uses public professional information only.
Current role and company
Karpathy works on pretraining at Anthropic, the AI company that makes Claude. Pretraining is the large-scale training stage that gives a model its core knowledge. VentureBeat reported that he will build a team aimed at using Claude to speed up pretraining research itself. Other outlets reported that he reports to Nick Joseph; VentureBeat did not state a reporting line, so treat that detail as reported rather than confirmed by Anthropic.
- Role (as reported)AI researcher on Anthropic's pretraining team; starting a team to use Claude to accelerate pretraining research.
- AnnouncedMay 19, 2026, in a post on X: "Personal update: I've joined Anthropic."
- Welcomed byNicholas Joseph, Anthropic's Head of Pretraining (on X, per VentureBeat).
- Eureka LabsKarpathy said he plans to resume his education work in time; VentureBeat infers the Eureka Labs work pauses meanwhile.
Career path
- 2005 to 2011: school. BSc at the University of Toronto (computer science and physics, minor in math), where he first met deep learning in Geoff Hinton's class, then an MSc at the University of British Columbia, per his own site.
- 2011 to 2015: Stanford PhD. Advised by Fei-Fei Li. He designed and taught CS231n, which his site calls Stanford's first deep learning course, and interned at Google Brain, Google Research and DeepMind.
- 2015 to 2017: OpenAI founding member. Research scientist and one of the company's original co-founders.
- 2017 to 2022: Tesla. Director of AI, leading the computer vision team for Autopilot and, briefly, Tesla Optimus, per his site. TechCrunch notes he left in 2022.
- 2023 to 2024: back to OpenAI. He built a team working on midtraining and synthetic data generation, per his site, and stepped down in February 2024.
- July 2024: Eureka Labs. TechCrunch reported he registered the company as a Delaware LLC on June 21, 2024 and announced it on X on July 16.
- May 2026: Anthropic. Joins the pretraining team.
What he is known for
- Teaching deep learning. CS231n at Stanford and, later, his YouTube series. His site lists a Zero to Hero series and talks on LLMs, in technical and general-audience tracks.
- Autopilot vision at Tesla. He led the computer vision team behind Autopilot and Full Self-Driving, according to coverage of his departure and hire.
- Eureka Labs. Eureka Labs describes itself as building a new kind of school that is AI native: teachers design the course materials and an AI Teaching Assistant guides students through them. Its first product is LLM101n, an undergraduate-level course on training your own AI. TechCrunch quoted his goal as letting anyone learn anything.
- Open-source projects. VentureBeat names Autoresearch, an LLM-driven tool that runs experiments in parallel, and the LLM Knowledge Base, a memory and context system for AI agents.
Leadership style and philosophy
Only his own words, as quoted by the outlets named. He has not given a long public account of how he runs teams, so this section is short.
- Back to research. In his announcement he said the next few years at the frontier of LLMs will be especially formative and that he is excited to get back to R&D (VentureBeat).
- Education still matters. He said he remains passionate about education and plans to return to it in time (VentureBeat, The Decoder).
- Open about changing his mind. The Decoder notes he said he was impressed by progress in agentic AI for coding after dismissing agentic capabilities a few months earlier.
- No drama in departures. TechCrunch reported he said there was no drama behind leaving Tesla or OpenAI.
Key numbers about Anthropic (company-reported)
Karpathy is a researcher, not a company executive, so these are company-wide figures from Anthropic's Series H announcement of May 28, 2026, nine days after he joined.
- Funding round$65 billion Series H, announced May 28, 2026.
- Post-money valuation$965 billion at the Series H.
- Run-rate revenueCrossed $47 billion in May 2026, per the announcement.
- Lead investorsAltimeter Capital, Dragoneer, Greenoaks and Sequoia Capital.
- Compute agreementsUp to 5 gigawatts of new capacity with Amazon and 5 gigawatts of next-generation TPU capacity with Google and Broadcom.
What to watch next
No papers, model releases or progress reports from his new team had been found when this page was written. Worth tracking: whether Anthropic describes the team's work publicly, and whether Eureka Labs and his open-source projects resume.
Questions people ask
Who is Andrej Karpathy?
He is an AI researcher who was a founding member of OpenAI, Tesla's Director of AI and the founder of Eureka Labs. He announced on May 19, 2026 that he joined Anthropic.
What does Karpathy do at Anthropic?
He works on the pretraining team and, per an Anthropic spokesperson quoted by VentureBeat, will start a team focused on using Claude to accelerate pretraining research.
Is Karpathy still at Eureka Labs?
He founded it in 2024 and said he plans to return to education work in time. VentureBeat infers the work pauses while he is at Anthropic. This page did not find a statement on the company's status.
Is he still at OpenAI or Tesla?
No. He left Tesla in 2022 and stepped down from OpenAI in February 2024, per TechCrunch.
Where did he study?
Per his site: BSc at the University of Toronto, MSc at the University of British Columbia and a PhD at Stanford under Fei-Fei Li.
Where can I learn from him?
His site lists a YouTube channel with the Zero to Hero series and talks on LLMs, and Eureka Labs' first course is LLM101n.
Sources
- Andrej Karpathy: personal site · Career history, education, YouTube series (his own page)
- VentureBeat: Andrej Karpathy announces he's joining Anthropic · May 19, 2026 announcement, role, quotes, open-source projects
- The Decoder: Karpathy picks Anthropic over former home OpenAI · Role, his comments on education and agentic coding
- Not a Tesla App: Former Tesla AI Director Andrej Karpathy Joins Anthropic · Tesla role, reporting to Nick Joseph
- TechCrunch: Karpathy's startup aims to apply AI assistants to education (July 16, 2024) · Eureka Labs founding, departures from Tesla and OpenAI
- Eureka Labs · Company description and LLM101n (company page)
- Anthropic: Series H announcement (May 28, 2026) · Company-reported funding, valuation, run-rate revenue
- Wikimedia Commons: Andrej karpathy 2016.webp · Hero photo, CC0, by Sk aman khan
Facts were read from the linked pages on October 31, 2026. Several pages were read through a fetch tool that returns summaries, so confirm details on the original page. Press reports are labelled as reports and company figures as company-reported. This page covers public professional information only.
Top photo: Sk aman khan / CC0 via Wikimedia Commons