Books.
The long way round, and the one that lasts. 7 of 7 are free to read online — posted by the authors or their publishers, not scanned by someone else.
Intuition first
Read these before the maths. They build the mental model that every book below quietly assumes you already have.
- Neural Networks and Deep Learning Michael Nielsen · 2015 The gentlest introduction that does not lie to you to stay gentle. Short, and it earns every page. Free · start here
- Dive into Deep Learning Zhang, Lipton, Li, Smola · 2023 Every idea comes with code you can run and break. This is one to work through at a keyboard, not read on a train. Free · core
The math
Enough linear algebra, probability and optimization to stop treating a model as a black box that either works or does not.
- Mathematics for Machine Learning Deisenroth, Faisal, Ong · 2020 Exactly the maths this field uses and not a chapter more. Start here if it is the notation, rather than the ideas, that keeps stopping you. Free · start here
- The Elements of Statistical Learning Hastie, Tibshirani, Friedman · 2009 The book everything else cites. Dense enough that reading it cover to cover is the wrong plan — come back to it with a question. Free · deep
- Deep Learning Goodfellow, Bengio, Courville · 2016 The standard graduate text. It predates the transformer era and is still the clearest account of what sits underneath it. Free · deep
Agents and systems
For the jump from a model that answers to a system that does something. Newer ground, and less settled.
- Principles of Building AI Agents Mastra · 2025 Architecture and implementation at book length, written by people who run this in production rather than demo it. Free · start here
- Reinforcement Learning: An Introduction Sutton & Barto · 2020 The canonical reinforcement learning text. Slower going than anything else here, and its ideas keep turning up in places you did not expect. Free · deep