🚗

Andrej Karpathy

Original nameAndrej Karpathy

Slovak-American Computer Scientist, Popularizer of AI Education

The Math Behind the Algorithm · Theoretical Founders

Technical Leadership of Tesla’s Self-Driving Vision TeamEarly Research at OpenAIAI Education Content Creation

Who they are

Andrej Karpathy is a computer scientist born in Slovakia and raised in Canada who studied for his Stanford doctorate under Fei-Fei Li, and in 2017 succeeded George Hotz as technical lead of Tesla’s self-driving vision team, directing the application of a pure-vision deep-learning system to the "Full Self-Driving" (FSD) function, a path choice that abandoned the LiDAR sensors peers commonly use, long controversial in the industry. Karpathy thereafter rejoined OpenAI for research, and is also one of the most influential content creators in AI education, his deep-learning-principle lecture-video series famed for a clear, accessible yet technically deep style, used by a great many students and practitioners worldwide as introductory learning material, and he often shares distinctive observations on AI trends on social media, including the recent concept of "vibe coding" describing a new paradigm of human-machine collaborative programming.

Primary sourcesKarpathy’s deep-learning teaching video series ("Neural Networks: Zero to Hero")

Key stories

Succeeding Hotz, Betting Tesla’s Self-Driving on "Pure Vision"

After taking over Tesla’s self-driving vision team, Karpathy continued and deepened the pure-vision technical path, abandoning the LiDAR sensors peers commonly use and holding that camera input alone with a strong enough neural network could likewise achieve reliable self-driving perception, a path choice long controversial in the industry, supporters holding it more in line with the natural logic of humans driving by vision alone, critics worried about the reliability risk of the pure-vision scheme in extreme weather or complex road conditions.

Relationships

Ally

🎙️Lex Fridman

Nemesis

🔧George Hotz　Karpathy succeeded Hotz as technical lead of Tesla’s self-driving vision team, and the two also had public disagreement on the choice of self-driving technical path

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

The proposal and popularity of the "vibe coding" conceptThe phrase "coding by feel" Karpathy proposed in recent years, describing the new programming paradigm in which developers rely more and more on large language models to help generate code rather than writing it line by line, was thereafter widely cited and discussed.

Appears in

AlphaGo, the Transformer, and the Race to ScaleThe Self-Driving Vision Dispute

← Back to Figures