Eureka Labs

Eureka Labs is an American artificial-intelligence-and-education company headquartered in San Francisco, founded in July 2024 by Andrej Karpathy, one of the most prominent AI researchers and educators globally. Karpathy was a founding member and research scientist at OpenAI (2015 to 2017), the Director of AI and Autopilot Vision at Tesla (2017 to 2022), the instructor of the influential Stanford CS231n course on convolutional neural networks (2015 onward), the founder of the Karpathy YouTube channel (over 700,000 subscribers as of April 2026, with the popular Zero-to-Hero educational series on neural networks built from scratch), and briefly returned to OpenAI as a senior research scientist (2023 to early 2024) before departing to start Eureka Labs. The company's stated thesis is that the post-2022 generative-AI wave makes AI-native educational technology — with AI tutors that scale teaching capacity at marginal cost — viable at quality levels that pre-AI educational technology could not match, and that the transition to AI-native education will be one of the principal commercial AI categories of the next decade. As of April 2026, Eureka Labs is one of the higher-profile AI-and-education startups globally, anchored on Karpathy's founder credibility and the inaugural LLM101n self-study course.

At a glance

• Founded: July 16, 2024, in San Francisco by Andrej Karpathy. Date corresponds to Karpathy's public announcement of the company.
• Status: Private. Seed-stage with continued early-stage funding.
• Funding: Approximately $20 million seed-stage capital reported in early-2025 industry coverage. Investors reportedly include Conviction (Sarah Guo), Sam Altman, and adjacent senior AI angel and venture investors.
• CEO: Andrej Karpathy, Founder and Chief Executive Officer. Slovak-Canadian-American AI researcher; PhD in computer science (Stanford under Fei-Fei Li, 2015); OpenAI founding member.
• Other notable leadership: Senior engineering and education-product leadership recruited through 2024 to 2025. Industry coverage has reported recruits from frontier-AI labs and adjacent edtech and AI organizations.
• Open weights: Yes, partial. The LLM101n self-study course is published openly through GitHub. Karpathy's broader educational content (YouTube channel, Zero-to-Hero series, lecture-by-lecture neural-network implementations) continues separately and predates Eureka Labs.
• Flagship outputs: LLM101n — the inaugural Eureka Labs course, a self-study large-language-model curriculum that walks students through building a Storyteller LLM from scratch in approximately 12 lectures. AI-native educational platform under continued development. Continued Karpathy YouTube and Zero-to-Hero educational content.

Origins

Eureka Labs was founded on July 16, 2024, in San Francisco by Andrej Karpathy, with the public announcement coinciding with Karpathy's departure from his second stint at OpenAI. Karpathy's founding statement framed Eureka Labs as an AI-native education company building a new kind of school — one where AI Teaching Assistants scale individual instruction in ways that human-teacher-only educational structures cannot match. The framing combined Karpathy's two principal career strengths: deep AI research credibility (OpenAI founding, Tesla autopilot, the convolutional-neural-network research lineage that has shaped much of modern computer vision) and educational track record (Stanford CS231n, the Karpathy YouTube channel and Zero-to-Hero series, which have together established Karpathy as one of the most-followed AI educators globally).

The founder cohort emerged narrower than typical Insurgent founding teams. Karpathy's public announcement framed the early-stage organization as a small team rather than a multi-co-founder organization, with continued recruiting expected through subsequent quarters. Early seed-stage capital was raised through 2024 to 2025, with industry coverage reporting approximately $20 million in cumulative seed funding from investors including Conviction (Sarah Guo's venture firm), Sam Altman, and adjacent senior AI angel and venture investors.

The inaugural product, LLM101n, was announced alongside the company founding as a self-study large-language-model course that walks students through building a "Storyteller LLM" from scratch in approximately 12 lectures. The course was framed as the first concrete output of Eureka Labs's broader AI-native educational platform — a deliverable demonstrating the educational-quality bar that the company aims to meet at scale. Through 2024 to 2026, LLM101n development has continued with selective public previews, and Karpathy's broader educational content (YouTube channel, Zero-to-Hero series, the November 2024 deep dive on LLMs, the February 2025 software 1.0/2.0/3.0 talk) has continued separately as personal-output that benefits Eureka Labs's brand visibility.

The 2024 to 2026 period has been substantially in early-stage build mode. Industry coverage has been comparatively quiet on Eureka Labs commercial-product specifics, with public-product launches limited to LLM101n and adjacent course content. The continued visibility of Karpathy's personal AI educational content has anchored Eureka Labs's brand visibility in ways that compensate for the early-stage commercial-product development cadence.

Mission and strategy

Eureka Labs's stated mission is to build a new kind of school that is AI-native — an educational organization where AI Teaching Assistants scale individual instruction at marginal cost while human teachers design and guide curricula at the strategy and adjustment levels. The strategy combines two threads. First, AI-native educational courses with the LLM101n inaugural offering and continued course-development pipeline anchored on AI Teaching Assistant infrastructure. Second, the underlying AI-tutor platform technology that supports the course-product line, with research and engineering investment that combines the Karpathy founder-credibility on AI research with edtech-product execution.

The competitive premise is that the post-2022 generative-AI wave has produced AI capabilities sufficient for AI Teaching Assistants to deliver tutoring quality that approaches one-on-one human tutoring at scale, that the educational-content market structurally rewards organizations that combine deep subject-matter expertise with engineering execution, and that Karpathy's combination of AI research credibility and educational track record provides a structurally distinctive founder positioning that horizontal-AI-platform alternatives and edtech-incumbent alternatives cannot match.

Models and products

• LLM101n. The inaugural Eureka Labs course. Self-study large-language-model curriculum walking students through building a Storyteller LLM from scratch in approximately 12 lectures. Public GitHub repository with continued content development.
• AI-native educational platform. Under continued development through 2024 to 2026. Public-product specifics have been limited.
• Karpathy YouTube channel and Zero-to-Hero series. Karpathy's personal educational output, predating and continuing alongside Eureka Labs. Over 700,000 subscribers as of April 2026.
• Karpathy LLM and software-3.0 explainer talks. Personal-output content anchoring Eureka Labs's brand visibility.

Distribution channels include open-source educational content distribution (GitHub for LLM101n, YouTube for Karpathy's personal content), with the AI-native educational platform commercial channel under continued development.

Benchmarks and standing

Eureka Labs's evaluation framework focuses on educational-content quality and adoption rather than horizontal foundation-model leaderboards. Karpathy's educational YouTube channel and Zero-to-Hero series have been consistently characterized as principal AI educational content globally, with substantial student and developer engagement. The LLM101n course has continued to attract educational-content adoption through 2024 to 2026 despite the in-development public-release cadence.

Industry coverage has consistently characterized Eureka Labs as one of the principal AI-and-education startups, with the Karpathy founder-credibility and the early-stage build-mode positioning as principal validating data points. Skeptical coverage has noted the comparatively limited commercial-product cadence relative to peer AI-and-education companies that have launched commercial offerings faster.

Leadership

As of April 2026, Eureka Labs's senior leadership includes:

• Andrej Karpathy, Founder and Chief Executive Officer.
• Senior engineering and education-product leadership recruited through 2024 to 2025.

The company has operated with a relatively small senior-leadership footprint consistent with its early-stage organizational structure.

Funding and backers

Approximately $20 million seed-stage capital reported in early-2025 industry coverage. Investors reportedly include Conviction (Sarah Guo), Sam Altman, and adjacent senior AI angel and venture investors. Specific cumulative private capital and current valuation are not publicly disclosed.

Industry position

Eureka Labs occupies a distinctive position as one of the principal AI-and-education startups globally, with founder credibility through Andrej Karpathy's AI research and educational track record, the inaugural LLM101n course, and the continued Karpathy personal-content brand visibility. Industry coverage has consistently characterized Eureka Labs as one of the structurally consequential AI-and-education ventures of the post-2022 wave, with the Karpathy founder-credibility as the principal validating data point.

The structural risks are two. First, the early-stage commercial-product cadence has been comparatively slow relative to peer AI-and-education organizations that have launched commercial offerings faster, and the gap between Karpathy's founder-credibility-implied expectations and the visible commercial-product output has been a recurring theme in industry coverage. Second, the AI-and-education category includes both incumbent edtech competitors (Khan Academy with Khanmigo, Duolingo with AI features, Coursera, edX) and frontier-AI-lab competitors (OpenAI ChatGPT educational integrations, Anthropic Claude educational integrations, Google AI Studio educational features) that bring substantial existing user-base distribution that Eureka Labs lacks.

Competitive landscape

• Khan Academy Khanmigo. Direct AI-and-education competitor with substantial existing student-user base. Different commercial structure (non-profit) but overlapping mission.
• Duolingo AI features. Adjacent AI-and-education competitor with substantial language-learning user base.
• OpenAI ChatGPT, Anthropic Claude. Frontier AI-lab educational integrations. Different commercial structure but the principal student-and-developer-facing AI tutoring alternatives in 2025 to 2026.
• Coursera, edX, Udacity. Incumbent edtech platforms with adjacent commercial positioning.
• fast.ai, Hugging Face educational content, the DeepLearning.ai Andrew Ng platform. Open-research AI educational peers.
• OpenAI, Tesla AI. Karpathy's prior employers; adjacent through founder-network and AI-research-community connections.

Outlook

• The continued AI-native educational technology product trajectory through 2026 to 2027.
• LLM101n public-release cadence and any expanded course offerings.
• Continued Karpathy personal-content output (YouTube, Zero-to-Hero, occasional deep-dive talks).
• Continued senior engineering and education-product recruitment.
• A potential Series A and the resulting commercial-product acceleration.

Sources

• Eureka Labs official site. Company reference.
• Andrej Karpathy Wikipedia. Founder reference.
• LLM101n on GitHub. Inaugural self-study course.
• Andrej Karpathy YouTube channel. Educational content.
• Karpathy Zero-to-Hero series. Educational content.
• Eureka Labs founding tweet (July 2024). Public announcement.