Andrej Karpathy earned a bachelor's degree at the University of Toronto (2005 to 2009), a master's at the University of British Columbia (2009 to 2011), and a PhD at Stanford University (2011 to 2015) under Fei-Fei Li. He was a founding research scientist at OpenAI, led the Autopilot computer vision team as Tesla's Director of AI from 2017 to 2022, and later moved toward teaching. The "Director of AI at Tesla" label in older headlines is therefore out of date.
Key takeaways
- Two of his three degrees come from Canadian universities, Toronto and UBC, and the third from Stanford.
- His own biography page gives the dates for each degree and names Fei-Fei Li as his PhD adviser.
- He designed and taught Stanford's CS231n; his page reports enrollment of 150 in 2015, 330 in 2016, and 750 in 2017.
- Tesla (2017 to 2022) is a past role, and his Eureka Labs school project was still described as under construction in the sources checked.
- A biography shows one person's path. It does not show that a given degree sequence or employer leads to similar results for others.
Degrees and where they were earned
Karpathy's personal biography page lists three degrees. At the University of Toronto he completed a bachelor's with a double major in computer science and physics and a minor in math. He writes that this is where he first got into deep learning, through Geoff Hinton's class and reading groups. At the University of British Columbia he worked with Michiel van de Panne on learning controllers for physically simulated figures. At Stanford he studied convolutional and recurrent neural networks for computer vision and language, with Fei-Fei Li of the Stanford Vision Lab as adviser.
The page also records internships at Google Brain (2011), Google Research (2013), and DeepMind (2015). Those are the only employer details it gives for the doctoral years, so nothing further is claimed here about their content.
Timeline from the biography page
The table restates dates as the person himself lists them. Roles and degree dates come from that one page, and a biography written by its subject can omit things.
| Years | Institution or employer | Role or credential |
|---|---|---|
| 2005 to 2009 | University of Toronto | Bachelor's, computer science and physics majors, math minor |
| 2009 to 2011 | University of British Columbia | Master's, learning controllers for simulated figures |
| 2011 to 2015 | Stanford University | PhD, adviser Fei-Fei Li |
| 2015 to 2017 | OpenAI | Research scientist and founding member |
| 2017 to 2022 | Tesla | Director of AI, Autopilot computer vision team, briefly Optimus |
| 2023 to 2024 | OpenAI | Returned to build a team on midtraining and synthetic data |
Teaching a large deep learning course
While a doctoral student, Karpathy designed Stanford's CS231n and was its primary instructor. His page calls it the first deep learning class at Stanford and reports enrollment growing from 150 students in 2015 to 330 in 2016 and 750 in 2017. Stanford's current CS231n course page is titled "Deep Learning for Computer Vision," lists offerings from Winter 2015 through Spring 2025, shows a Spring 2026 offering, and does not name him among its current instructors. The course has continued under other teaching teams.
The enrollment figures are self-reported and were not confirmed against a registrar source. They are useful mainly as a sign that an introductory neural network course drew steadily more students within three years.
From Tesla to a teaching project
The Tesla chapter ended in 2022, according to his page, and he returned to OpenAI in 2023. Since 2024 his page describes educational AI videos on his YouTube channel. Separately, an announcement dated July 16, 2024 introduced Eureka Labs, which describes itself as a school built around a human teacher who designs course materials and an AI teaching assistant that guides students through them. Its first planned course, LLM101n, is described as an undergraduate-level class on training a language model.
Status matters here. The public LLM101n repository on GitHub was archived by its owner on August 1, 2024, and its README states that the course does not yet exist and is being developed by Eureka Labs. The outline lists 17 chapters, from a bigram model and backpropagation to attention, fine-tuning, and deployment. Whether the course has launched could not be confirmed, so it should be treated as a stated plan and not as an available program.
Reading a famous person's education record
Stories about well-known researchers tend to turn degree lists into recipes. A few limits apply to this one:
- Timing mattered. Deep learning was moving quickly between 2011 and 2015, and the biography does not isolate how much of his later position came from the degrees.
- Advisers and classmates shaped the work as much as the schools did, and the page names several people he worked with.
- One path says nothing about the average graduate of the same programs.
- Employer titles change. A current profile should be checked before a role is quoted.
For students in Canada, the narrower and safer point is that undergraduate and master's training at public universities in Toronto and Vancouver appears in this particular record.
FAQ
Where did Andrej Karpathy study?
According to his biography page, he completed a bachelor's at the University of Toronto, a master's at the University of British Columbia, and a PhD at Stanford University.
Is he still Tesla's Director of AI?
No. His page lists Tesla from 2017 to 2022. He returned to OpenAI in 2023 and the page describes his work since 2024 as educational AI videos.
Did he teach at Stanford?
He designed CS231n and was its primary instructor while a doctoral student, by his own account. The course page today lists other offerings and teaching staff.
Is Eureka Labs an operating school?
That could not be confirmed. The public repository for its first course says the course does not yet exist, and the sources checked describe the school as being built.
Related reading
- Other leading AI researchers and what they work on
- How Tesla uses AI in its vehicles
- AI technology in education
- Potential and limits of AI in classrooms
More in AI in education.
Last verified: October 2026
Sources
- Andrej Karpathy, personal biography page, accessed October 2026
- CS231n: Deep Learning for Computer Vision, Stanford University, 2026
- LLM101n repository and README (archived August 1, 2024), GitHub
- Eureka Labs announcement page (July 16, 2024)