Andrej Karpathy

Early Life and Education in Canada

Andrej Karpathy was born on 23 October 1986 in Bratislava, the capital of what was then Czechoslovakia and is now Slovakia. His family relocated to Canada when he was 15 years old, settling in the city of Toronto, Ontario. This move integrated him into the Canadian education system during his formative high school years. He pursued post-secondary education at the University of Toronto, where he undertook a demanding dual-degree program, culminating in bachelor's degrees in both Computer Science and Physics in 2009. His time at U of T provided a strong technical foundation in both theoretical and applied sciences.

For his graduate studies, Karpathy moved to the west coast to attend the University of British Columbia in Vancouver. He completed his master's degree there in 2011, conducting research on physically simulated figures, which involves creating digital models that obey physical laws. Concurrently with his formal studies, he cultivated an early online presence as an educator. In 2006, he started a YouTube channel under the username badmephisto, which initially focused on tutorials for solving the Rubik's Cube. The channel's influence was notable, as world champion speedcuber Feliks Zemdegs publicly acknowledged learning from Karpathy's videos. This activity demonstrated an early aptitude for explaining complex systems in an understandable way.

Doctoral Research and Foundational Teaching at Stanford

After his studies in Canada, Andrej Karpathy entered the PhD program in computer science at Stanford University in California. He conducted his doctoral research under the supervision of Professor Fei-Fei Li, a leading figure in computer vision. His 2015 dissertation focused on developing deep learning models for tasks that combine natural language processing and computer vision, a challenging area known as multimodal learning. This work required creating neural networks capable of understanding and generating connections between textual descriptions and visual content, research that directly informed later advances in large-scale AI models.

During his PhD studies, Karpathy was tasked with developing a new course on an emerging topic. He authored and became the primary instructor for CS 231n: Convolutional Neural Networks for Visual Recognition, which was Stanford's first course dedicated entirely to deep learning. The course launched in the 2015 academic year to an initial enrolment of 150 students. Due to exploding interest in the field, the class size expanded rapidly, reaching 750 students by 2017 and becoming one of the largest courses at the university. The course materials, including detailed lecture notes, video recordings, and programming assignments, were made publicly available online and became a standard reference for thousands of students and professionals worldwide seeking to understand convolutional neural networks.

Industry Career: OpenAI and Tesla

Upon receiving his PhD in 2015, Andrej Karpathy became a founding member of the artificial intelligence research organization OpenAI. He worked there as a research scientist from 2015 until 2017, contributing to the lab's early projects in deep learning and artificial general intelligence research. In June 2017, he transitioned to the automotive sector, accepting the position of director of artificial intelligence at Tesla. In this role, he reported directly to the company's chief executive, Elon Musk, and held leadership responsibility for the Autopilot Vision team. His work involved overseeing the development of the computer vision systems that interpret camera data for Tesla's driver-assistance and autonomous driving features.

Karpathy worked at Tesla for nearly five years. In early 2022, he took a sabbatical from the company that lasted several months. He formally announced his departure from Tesla in July 2022. Following his exit, he dedicated time to independent projects and education, producing a series of YouTube videos that detailed the process of building artificial neural networks from first principles. In February 2023, he announced his return to OpenAI, rejoining the company for a second period of employment. This tenure lasted exactly one year, as an OpenAI spokesperson confirmed his departure on 13 February 2024. His industry contributions were recognized in 2020 when MIT Technology Review named him one of its Innovators Under 35.

• 2015: Completes PhD and co-founds OpenAI as a research scientist.
• June 2017: Leaves OpenAI to join Tesla as Director of Artificial Intelligence.
• 2020: Named an MIT Technology Review Innovator Under 35.
• Early 2022: Begins a several-months-long sabbatical from Tesla.
• July 2022: Announces his departure from Tesla.
• February 2023: Announces return to OpenAI.
• February 2024: Departs OpenAI confirmed by company spokesperson.

Founding Eureka Labs and AI Education Advocacy

In 2024, Andrej Karpathy founded a new company focused on artificial intelligence education, named Eureka Labs. He announced the launch on 16 July 2024 through a post on his X.com account. The company's stated mission is to develop and provide high-quality educational resources for understanding and working with artificial intelligence, with a specific emphasis on large language models. Eureka Labs released its inaugural product, an online course titled LLM101n, which offers structured lessons on the fundamentals of LLMs. The company maintains an official website and an official GitHub account where it shares code and supplementary educational materials.

Alongside the formal LLM101n course, Karpathy continues a broader instructional effort he calls the Zero to Hero series, which consists of free tutorials covering foundational concepts of large language models. A distinctive and debated aspect of Eureka Labs' philosophy is its active advocacy for deploying AI teaching assistants within educational environments. This position has attracted criticism from some educators and privacy experts, who cite concerns over student data privacy and the potential weakening of the direct interpersonal connection between teachers and students. In the same year he founded the company, Karpathy's influence was acknowledged by Time Magazine, which included him in its 2024 list of the 100 Most Influential People in AI.

Coining 'Vibe Coding' and Joining Anthropic

In the period between founding Eureka Labs and his next career move, Andrej Karpathy contributed a new term to the discourse around AI-assisted development. In February 2025, he coined the phrase vibe coding to describe a method of software creation enabled by advanced AI tools. Vibe coding refers to the practice where hobbyists and developers can build functional applications and websites primarily by typing descriptive, natural language prompts into AI coding assistants, rather than manually writing extensive lines of traditional code. This concept highlights a shift towards more intuitive and accessible forms of programming, though it also prompts discussion about the evolving skill set required for software engineering.

On 19 May 2026, Karpathy announced a significant career transition via a statement posted on X.com. He revealed that he had joined Anthropic, an AI safety and research company known for its work on constitutional AI. Anthropic confirmed the hiring, stating that Karpathy would be leading a team dedicated to research in pretraining. Pretraining is the initial, computationally intensive phase of developing large language models, where a model learns from a vast corpus of text data before any specific task fine-tuning. This role places him at the centre of fundamental model development research within an organization that prioritizes AI alignment and safety.

Teaching Methodology and Public Engagement

Andrej Karpathy's approach to teaching is characterized by a clear, systematic breakdown of complex topics combined with an emphasis on practical implementation. His instructional style, evident in his Stanford course and his online tutorials, involves explaining detailed subjects like convolutional neural networks and transformer architectures step by step. He strongly advocates for learning by building, a method he demonstrated in his YouTube series after leaving Tesla, where he coded neural networks from scratch with detailed commentary. This philosophy of understanding through creation connects his academic teaching, his industry blog posts, and the commercial courses offered by Eureka Labs.

His public engagement extends across several platforms where he shares technical insights and educational content. Karpathy maintains a personal website and an active blog, which includes a version hosted on GitHub that allows for open feedback and collaboration. He continues to post videos to his badmephisto YouTube channel, though the content has evolved from Rubik's Cube tutorials to in-depth AI explanations. This combination of university instruction, open-source online tutorials, and structured commercial courses represents a comprehensive strategy for educating people about artificial intelligence. His consistent output has established him as a primary resource for learners worldwide seeking to understand advanced AI systems.

• Maintains a personal website and a blog (including a GitHub version).
• Runs the YouTube channel 'badmephisto', started in 2006.
• Authored and taught Stanford's CS 231n, growing it from 150 to 750 students.
• Creates the 'Zero to Hero' free tutorial series on LLM fundamentals.
• Developed the commercial course 'LLM101n' for Eureka Labs.
• Advocates for the use of AI teaching assistants in education.

Recognition and Influence

Andrej Karpathy's work has received formal recognition from major publications and institutions within the technology sector. In 2020, MIT Technology Review selected him for its annual Innovators Under 35 list, an award that highlights promising young technologists. Four years later, in 2024, Time Magazine included him in its inaugural Time100 AI list, naming him one of the 100 Most Influential People in AI. This recognition came in the same year he founded Eureka Labs and departed OpenAI, which highlights his continued relevance in the field. These honours reflect his impact across research, industry application, and education.

His influence is perhaps most tangibly measured by the widespread adoption of his educational materials. The lecture videos and assignments from his Stanford CS 231n course have been viewed and used by hundreds of thousands of individuals beyond the university's walls. His online writings, such as blog posts on neural network architectures, are frequently cited and discussed within AI communities. The term he coined, vibe coding, has entered the lexicon of developers discussing AI-powered programming tools. Through these channels, Karpathy has shaped the learning pathways for countless students and professionals entering the field of artificial intelligence, establishing a durable educational legacy.

Questions about Andrej Karpathy

Where did Andrej Karpathy go to university in Canada?

Andrej Karpathy completed his undergraduate studies in Canada at the University of Toronto, where he earned dual bachelor's degrees in Computer Science and Physics in 2009. He then pursued a master's degree at the University of British Columbia in Vancouver, graduating in 2011. His research at UBC focused on physically simulated figures. These Canadian institutions provided the foundational education that preceded his doctoral work at Stanford University in the United States.

What is Andrej Karpathy's connection to Tesla?

Andrej Karpathy joined Tesla in June 2017 as the director of artificial intelligence, reporting directly to CEO Elon Musk. In this role, he led the Autopilot Vision team, which was responsible for developing the computer vision systems crucial for Tesla's autonomous driving features. He took a sabbatical in early 2022 and subsequently announced his departure from the company in July 2022. His nearly five-year tenure was a significant period in the development of Tesla's AI-driven automotive technology.

What is Eureka Labs?

Eureka Labs is an AI education company founded by Andrej Karpathy in July 2024. The company's goal is to create accessible and high-quality educational resources for learning about artificial intelligence, with a particular focus on large language models. Its first product was the AI course LLM101n. The company also promotes the Zero to Hero tutorial series and advocates for the use of AI teaching assistants in education, a stance that has generated some debate regarding data privacy and student-teacher interaction.

What did Andrej Karpathy do at OpenAI?

Andrej Karpathy was a founding member and research scientist at OpenAI from 2015 to 2017, contributing to the organization's early deep learning research. He returned to OpenAI in February 2023 for a second tenure, but left again in February 2024. His work at OpenAI involved research in artificial intelligence during key formative years for the company. His departures were confirmed by company spokespersons, with his 2024 exit coinciding with the year he was named to Time Magazine's list of the 100 Most Influential People in AI.

What is Andrej Karpathy's role at Anthropic?

Andrej Karpathy joined Anthropic in May 2026 to lead a team for research in pretraining. He announced this move via a statement on X.com. This role followed his work at other major AI organisations. Prior to this, he co-founded and worked at OpenAI, served as Tesla's director of artificial intelligence and Autopilot Vision, and founded the AI education platform Eureka Labs in 2024. His career has consistently focused on deep learning and artificial intelligence research and development.

What is Andrej Karpathy known for teaching?

Andrej Karpathy is best known for creating and teaching Stanford University's first deep learning course, CS 231n: Convolutional Neural Networks for Visual Recognition. He authored the course and served as its primary instructor, overseeing its growth from 150 students in 2015 to 750 students by 2017. He also produces extensive online educational content, including his early badmephisto YouTube channel which featured Rubik's Cube tutorials. Through his company Eureka Labs, he created the AI course LLM101n and a broader educational series called Zero to Hero on LLM fundamentals.

Sources and further reading

• Andrej Karpathy (Wikipedia)
• Andrej Karpathy Explained: Career, Vibe Coding & More (mrprompts.substack.com)
• Andrej Karpathy, AGI is still a decade away (dwarkesh.com)

Checked by the Your Space Hamilton desk. Last updated 2026-10-18.