d2l-ai/d2l-en

D2L.ai, short for "Dive into Deep Learning," is a free, open-source textbook designed to teach deep learning (the branch of AI that powers things like image recognition, chatbots, and recommendations) by combining explanations, math, and runnable code all in one place. The idea is that the best way to understand these concepts is to actually run and tweak the code yourself, not just read theory. The book lives as a collection of Jupyter notebooks, interactive documents where you can read the explanation and execute the code in the same window. Each chapter covers a concept, then shows you working code examples you can run immediately without any extra setup. Who would use this? Anyone who wants to go from "I've heard of AI" to "I can actually build and train models", whether you're a student, a self-taught developer, a PM wanting to understand the technology, or a founder evaluating what's technically possible. It assumes some math familiarity but no prior AI experience. It's been adopted by hundreds of universities including Stanford, MIT, Harvard, and Cambridge, which signals that it's structured enough for formal coursework but accessible enough for self-study. The codebase and content are written in Python, and the notebooks are designed to run on multiple deep-learning frameworks so you're not locked into one ecosystem. Topics include computer vision (teaching computers to see), data science fundamentals, and advanced techniques like Gaussian processes and hyperparameter optimization. The book is free under a Creative Commons license, and there's an active community forum for questions.