Practical ML for Computer Vision: I read it to be able to say no I do not sell computer vision work. I read this so I could explain, with arguments, why most requests that begin with "just take a photo" end badly. AI 1 jun 2026
Mathematics for Machine Learning: reading it so I do not have to take anyone's word It does not teach machine learning: it teaches the mathematics underneath, and the authors warn you in the preface. I read it so I could argue with vendors instead of believing them. AI 13 may 2026
Machine Learning with PyTorch and Scikit-Learn: the rare class decides everything I have the first part of this book, the classical one. I will not opine on the neural network half because I did not read it. Of what I did read, the evaluation chapter is the one I use to say no. AI 28 abr 2026
Learning LangChain: the graph survives, the code does not A library book, and library books age at the speed of the library. What does survive is a way of thinking about AI workflows that I have not dropped since. AI 19 abr 2026
Large Language Models in Finance: two chapters that earn it, and a lot around them Written for banks and quantitative finance, and it sprawls over everything it passes. But it holds two control ideas I carried whole into work that has nothing to do with finance. AI 9 abr 2026
LLM Engineer's Handbook: the shape of the system beats the model A book that builds a whole system around a language model, from data collection to deployment. I took the architecture; I left the fine-tuning behind. AI 28 mar 2026
Human-Centered AI: a summer school bound as a book Twenty lectures from a European advanced course, published as a volume. Half of it is a neural network course you can get better elsewhere; the other half earns the book. AI 16 mar 2026
Hands-On Machine Learning: chapter two is worth half the book A machine learning manual with code. The first half is still a solid apprenticeship; the second half aged badly, and that is worth saying out loud. AI 13 mar 2026
Hands-On Large Language Models: half the AI requests do not need a generator The best-illustrated book in the field, and the illustrations are not decoration, they are the content. It left me the criterion that has saved the people I work with the most money. AI 9 mar 2026
Hands-On Generative AI: an early draft, with the good and bad of being one An unedited early release on using and fine-tuning open models for text, image and audio. Good for understanding the pieces; not good for learning in order. AI 6 mar 2026
Generative Deep Learning: why the same question does not give the same answer A tour of the families of generative models with code that runs on an ordinary machine. It gave me the answer to the question that comes up in every meeting: why did it say something different yesterday. AI 28 feb 2026
Generative AI Design Patterns: thirty-two names for things already being done A numbered catalogue of patterns for language-model applications. I read the hundred-page excerpt rather than the full book, and I say so because it changes what I can claim. AI 25 feb 2026