Andrej Karpathy is one of the most influential figures in modern artificial intelligence, a founding member of OpenAI, the former director of AI and Autopilot Vision at Tesla, and the founder of the AI-education startup Eureka Labs. In May 2026 he made headlines by joining rival lab Anthropic, the maker of Claude, to work on the pre-training team behind its frontier models. Along the way he coined the term “vibe coding,” wrote the widely cited essay “Software 2.0,” and built some of the most-watched AI courses on the internet.

This guide covers who Andrej Karpathy is, his full career timeline from OpenAI to Tesla and back, why his move to Anthropic matters, and the ideas that made him famous. You will also find his recent open-source work like nanochat and autoresearch, his views on AI agents, and straight answers on his age, education, and what is actually known about his net worth.

The Key Takeaways

  • Born October 23, 1986 in Bratislava; a Slovak-born AI researcher who moved to Canada at 15, now 40.
  • Founding member of OpenAI (2015–2017) and Tesla’s Director of AI (2017–2022), where he led the Autopilot computer vision team.
  • Founded Eureka Labs in July 2024 and joined Anthropic in May 2026 on its pre-training team; he says he will return to education “in time.”
  • Coined “vibe coding” and authored the influential “Software 2.0” essay.
  • He has never disclosed his net worth, and no official source states one.

Who Is Andrej Karpathy?

Del editor

Descubre qué IA responde mejor

Hazles la misma pregunta a GPT, Claude y Gemini y compara las respuestas. En Mac, iPhone y iPad.

¡Descárgala ahora!
Del editor

Descubre qué IA responde mejor

Hazles la misma pregunta a GPT, Claude y Gemini y compara las respuestas. En Mac, iPhone y iPad.

Pruébala gratis
Del editor

Deja de pagar tres suscripciones de IA

GPT, Claude y Gemini en una sola app y con un solo plan. En Mac, iPhone y iPad.

¡Descárgala ahora!
Del editor

Deja de pagar tres suscripciones de IA

GPT, Claude y Gemini en una sola app y con un solo plan. En Mac, iPhone y iPad.

Pruébala gratis
Del editor

Deja de saltar entre apps de IA

Hazle la misma pregunta a ChatGPT, Claude y Gemini en una sola ventana y quédate con la mejor respuesta. En Mac, iPhone y iPad.

Pruébala gratis
Del editor

Deja de saltar entre apps de IA

Hazle la misma pregunta a ChatGPT, Claude y Gemini en una sola ventana y quédate con la mejor respuesta. En Mac, iPhone y iPad.

¡Descárgala ahora!
Del editor

No solo leas sobre modelos de IA: pruébalos

Los modelos de IA más recientes, listos en una sola app. En Mac, iPhone y iPad.

¡Descárgala ahora!
Del editor

No solo leas sobre modelos de IA: pruébalos

Los modelos de IA más recientes, listos en una sola app. En Mac, iPhone y iPad.

Pruébala gratis

Andrej Karpathy is an AI researcher born on October 23, 1986 in Bratislava, then part of Czechoslovakia, which makes him 40 years old as of November 2026. His family moved to Toronto when he was 15. He is best known for his work in deep learning and computer vision, and for explaining hard AI ideas in a way ordinary people can follow. TIME named him to its TIME100 Most Influential People in AI list in 2024, and his tutorials reach millions of developers.

His academic path runs through three countries. He earned a bachelor’s degree in computer science and physics at the University of Toronto, a master’s at the University of British Columbia, and a PhD from Stanford under Fei-Fei Li, focusing on the intersection of computer vision and natural language. While at Stanford he created and taught CS231n, the university’s first deep learning course and still one of its most popular.

Andrej Karpathy’s Career Timeline

Karpathy has moved between the biggest names in AI, often at pivotal moments. Here is the full path from OpenAI to Tesla and on to Anthropic.

Years Role Organization Focus
2015–2017 Founding member OpenAI Deep learning, computer vision research
2017–2022 Director of AI / Autopilot Vision Tesla Full Self-Driving and Autopilot
2023–2024 Researcher OpenAI (return) Midtraining and synthetic data
2024–present Founder Eureka Labs AI-native education (on hold)
May 2026–present Researcher Anthropic Pre-training for Claude

Karpathy first joined OpenAI as a founding member in 2015, then left in 2017 to lead AI at Tesla, where he ran the Autopilot Vision team and the company’s Full Self-Driving effort. He stepped away from Tesla in 2022, returned to OpenAI for about a year, and then struck out on his own. The through-line in his career is a focus on teaching machines to see and, increasingly, on teaching people to build AI.

Why Did Andrej Karpathy Leave Tesla and OpenAI?

Karpathy left Tesla in July 2022 after five years, calling it “a difficult decision to part ways” and noting that in that time Autopilot “graduated from lane keeping to city streets.” He rejoined OpenAI in February 2023 and left again in February 2024. His own explanation was blunt: “nothing ‘happened’ and it’s not a result of any particular event, issue or drama.” His plan, he wrote, was to work on personal projects, which became Eureka Labs five months later.

What Is Eureka Labs?

Eureka Labs is the AI-education company Karpathy announced on July 16, 2024, built around the idea of an “AI-native” school where an AI teaching assistant guides students through course material. Its planned first course, LLM101n, is meant to walk learners through building their own small language model, a “Storyteller AI,” from scratch using Python, C, and CUDA. As of November 2026 the course has not launched: its GitHub page still says it “does not yet exist,” and the repository has been archived since August 2024. When he joined Anthropic, Karpathy said he plans to resume his education work “in time.”

Why Did Andrej Karpathy Join Anthropic?

On May 19, 2026, Karpathy announced he had joined Anthropic, the company behind Claude, sending a jolt through the AI world given his OpenAI roots. He is working on the pre-training team under team lead Nick Joseph, the group responsible for the massive training runs that give Claude its core knowledge and capabilities. Anthropic told TechCrunch he would start a team focused on using Claude itself to accelerate pre-training research.

The optics are hard to miss. A founding member of OpenAI now helps train its fiercest rival’s models.

In his own words, Karpathy said he believes “the next few years at the frontier of LLMs will be especially formative” and that he was excited to get back to hands-on research and development. The move reads as a vote of confidence in Anthropic’s research direction, and it lands amid an intense talent war between the top labs. If you want the bigger picture of how these two companies stack up, see our breakdown of Anthropic vs OpenAI.

What Is Vibe Coding? Karpathy’s Big Ideas

Vibe coding, a term Karpathy coined in a February 2025 post on X, describes building software by telling an AI model what you want in plain language and letting it write most of the code. Instead of hand-writing every line, you guide and steer the model, leaning on its output and your own judgement. The phrase caught on fast and now shows up everywhere from startup demos to job descriptions. If you want to try the workflow yourself, our guide to the best free AI for coding is a good place to start.

His other landmark idea is “Software 2.0,” a November 2017 essay arguing that neural networks represent a new kind of programming. In this view the “code” is no longer written by hand but learned, with a model’s trained weights doing the work that explicit instructions used to do. The framing became a standard reference for how the industry thinks about machine learning, and it still shapes how engineers describe what AI is doing to their craft.

Karpathy’s Recent Work: nanochat, Autoresearch, LLM Knowledge Bases, and HackerTube

Karpathy stays famous partly because he keeps shipping. His project nanochat, billed as “the best ChatGPT that $100 can buy,” is a minimal, readable codebase that trains a small chat model end to end so people can see exactly how the pieces fit together. In March 2026 he released autoresearch, a roughly 630-line tool that lets AI agents run their own machine-learning experiments on a single GPU. Left running for about two days, the agent found around 20 changes that improved the model and cut nanochat’s “Time to GPT-2” leaderboard score from 2.02 to 1.80 hours, about 11 percent. By November 2026 the repository had passed 100,000 GitHub stars.

In April 2026 he sparked another wave with a post on “LLM Knowledge Bases,” viewed more than 22 million times by November 2026. He described feeding articles, papers, and repos into a folder and having a model “compile” them into an interlinked wiki of Markdown files, with summaries, backlinks, and concept pages. He wrote that a large fraction of his own AI usage was going “less into manipulating code, and more into manipulating knowledge.” To go deeper on building skills like this, see our guide on how to learn AI.

At the end of October 2026 came HackerTube, a free site that gathers the 6,454 YouTube videos Hacker News commenters recommended between 2007 and 2026, sorted into 150 categories. Its credits say Claude, working in Claude Code, built the data pipeline and the site, and that Karpathy “prompted [it] into existence.” It is a small, public demo of the agent-driven workflow he keeps writing about.

Why Karpathy Says AI Agents Need a “Decade”

Despite the hype around autonomous AI, Karpathy is a measured voice. On the Dwarkesh Podcast in October 2025 he argued that fully working AI agents will take about a decade to arrive, because today’s systems lack enough intelligence, multimodality, and continual learning to be trusted with real tasks. In a December 2025 post on X he admitted he had “never felt this much behind as a programmer,” adding that the profession “is being dramatically refactored” by agents, prompts, and new tooling. His point is that the job is changing fast, not that engineers are obsolete.

Andrej Karpathy’s Net Worth and Personal Life

Karpathy has never publicly disclosed his finances, and no verified net worth figure exists. The numbers on net-worth sites are guesses that vary widely and cite no filing or disclosure. What is on the record is the career behind any wealth: senior roles at OpenAI and Tesla, his own company Eureka Labs, and since May 2026 a research job at Anthropic.

He keeps his private life out of the spotlight and does not discuss a wife or partner publicly. What he shares openly is his work, most of it free, from YouTube deep-dives like building a GPT from scratch to open-source repositories anyone can study. That generosity with knowledge is a big reason his name carries so much weight in AI.

The Bottom Line

Andrej Karpathy matters not just for where he has worked, OpenAI, Tesla, and now Anthropic, but for how he has shaped the way people understand AI. He turns frontier research into ideas and tools that ordinary developers can actually use, and his move to Anthropic signals where he thinks the next breakthroughs will come from. To see the founders building alongside him, read our profiles of Sam Altman and Anthropic CEO Dario Amodei, plus co-founder Daniela Amodei, former OpenAI CTO Mira Murati, alongside DeepMind co-founder Demis Hassabis on the Google side of the race, plus DeepSeek founder Liang Wenfeng driving the open-source push from China. For the head of Microsoft AI, see Mustafa Suleyman.

You do not need to pick a single lab to benefit from their work. With Fello AI you can use Claude, ChatGPT, Gemini, Grok, and DeepSeek side by side on your Mac, iPhone, or iPad for one monthly price, so you always have the best model for the task in front of you.

FAQ

Is Andrej Karpathy a co-founder of OpenAI?

Yes. Karpathy was a founding member of OpenAI in 2015, left in 2017 for Tesla, returned from 2023 to 2024, and joined Anthropic in May 2026.

What is Andrej Karpathy doing now?

Since May 2026 he has worked on Anthropic’s pre-training team, the group that trains the Claude models. He has said he plans to resume his education work “in time,” and he still ships side projects such as HackerTube.

What is vibe coding?

Vibe coding is a term Karpathy coined in February 2025 for building software by describing what you want in natural language and letting an AI model write most of the code while you guide it.

How much is Andrej Karpathy worth?

No verified figure exists. He has never disclosed his net worth, so any number you see online is an estimate, not data.

Where is Andrej Karpathy from?

He was born in Bratislava, then part of Czechoslovakia and now the capital of Slovakia, in 1986. His family moved to Toronto when he was 15, and he studied in Canada and the United States.

La imagen destacada de este artículo se creó con ayuda de la IA.