What are the best books on AI engineering?

Key facts

• Chip Huyen's AI Engineering (O'Reilly, December 2024) covers building applications on foundation models, which involves more prompt engineering, context construction and parameter-efficient fine-tuning. (Source: Chip Huyen, AI Engineering book repository)
• Huyen's Designing Machine Learning Systems (O'Reilly, May 2022) covers applications built on traditional ML models, which involve more data annotation, feature engineering and model training. (Source: Chip Huyen, books)
• Sebastian Raschka's Build a Large Language Model (From Scratch), published by Manning in September 2024, codes a GPT-style model without existing LLM libraries, then pretrains and fine-tunes it. (Source: Manning Publications)
• The second edition of Designing Data-Intensive Applications, by Martin Kleppmann and Chris Riccomini, was published by O'Reilly in March 2026. (Source: Martin Kleppmann, Designing Data-Intensive Applications, second edition)
• Reliable Machine Learning (O'Reilly, 2022), by Cathy Chen, Niall Richard Murphy, Kranti Parisa, D. Sculley and Todd Underwood, applies site reliability engineering principles to machine learning in production. (Source: Google Research: Reliable Machine Learning)
• Aurélien Géron's Hands-On Machine Learning with Scikit-Learn and PyTorch (O'Reilly, 2025) is a new PyTorch-based book from the author of the three TensorFlow-based editions of Hands-On Machine Learning. (Source: Aurélien Géron, handson-mlp repository)
• From Model to Product (independently published, March 2026) is written by Dr. Dulan Dias, the author of this site. (Source: From Model to Product, book page)

Definitions

LLM application

Software whose behaviour depends on a large language model used through prompts, retrieved context or fine-tuning, plus the code, data and controls around it.

ML systems design

Deciding how data, features, models, training, serving and monitoring fit together so that a machine learning system meets its requirements in production, not just on a test set.

MLOps

The practices that join building ML systems to operating them: automated pipelines, versioning of code, data and models, testing, deployment, monitoring and retraining.

How these books were chosen

Each book here still holds up in November 2026, is written by people who have built or run the systems they describe, and earns its place for a specific need. Titles, authors, publishers and years were checked against the publisher's or the author's own page. Books that mainly document one framework's API were left out, because they date within months.

Disclosure: From Model to Product is written by Dr. Dulan Dias, who also writes this site. It appears in the group where it belongs, with an account of where it differs, and is not ranked above the others.

The list at a glance

Foundations: how models and LLMs work

Hands-On Machine Learning with Scikit-Learn and PyTorch, by Aurélien Géron, is the practical route into machine learning for programmers. It runs from regression and an end-to-end Scikit-Learn project to neural networks, transformers and diffusion models in PyTorch, fine-tuning pretrained models and reinforcement learning. Read it if overfitting, validation sets or precision versus recall are still fuzzy: you need them to judge any model, an LLM included.

Build a Large Language Model (From Scratch), by Sebastian Raschka, builds a GPT-style model step by step: preparing text data, coding attention, pretraining, then fine-tuning for classification and for following instructions. You will not train foundation models at work, but after this book tokenisation, context windows and fine-tuning stop being black boxes.

Building LLM applications

AI Engineering, by Chip Huyen, is the most complete single book on building with foundation models. It covers whether to build at all, evaluation (including using AI to judge AI), prompt engineering, RAG, agents, fine-tuning, dataset engineering and making models faster and cheaper. It is deliberately not a tutorial and has little code, which is why it stays useful after the APIs it mentions change.

Hands-On Large Language Models, by Jay Alammar and Maarten Grootendorst, teaches through almost 300 custom figures and runnable notebooks: tokens and embeddings, the inside of a transformer, classification and clustering, prompt engineering, semantic search and RAG, multimodal models and fine-tuning. It suits engineers who learn visually and want to run every example.

LLM Engineer's Handbook, by Paul Iusztin and Maxime Labonne, takes one project the whole way: data collection, a training pipeline, a simple RAG system, deployment on AWS, monitoring, and testing and evaluation. Read it to see every stage wired together in code, and expect its tool choices to date faster than its structure.

ML systems design

Designing Machine Learning Systems, by Chip Huyen, treats an ML system as a chain of decisions about training data, features, how often to retrain and what to monitor, with case studies. Its examples are classical ML, but the material on distribution shift, monitoring and continual learning applies directly to LLM products. Huyen positions AI Engineering as its companion: one centres on traditional models, the other on foundation models.

Designing Data-Intensive Applications is not an AI book, and it belongs on the list anyway, because an AI product is a data system first. Replication, partitioning, transactions, streams and the trade-offs between storage engines decide whether retrieval, feature serving and logging hold up. The 2026 second edition, by Martin Kleppmann and Chris Riccomini, updates the first for current technologies and cloud services.

Production and MLOps

Reliable Machine Learning, by Cathy Chen, Niall Richard Murphy, Kranti Parisa, D. Sculley and Todd Underwood, applies site reliability engineering to machine learning, from monitoring models in production to how an organisation should run model development. It suits engineers who will be on call for a model and the leads who build the team around it.

From Model to Product, by Dr. Dulan Dias (the author of this site), covers the system around a model: problem framing, choosing between rules, classical ML and LLMs, data and storage architecture, APIs, deployment, monitoring, reliability, cost, teams and governance. Where it differs: it carries one product, a waste-sorting assistant that combines an on-device vision model, rules and a language model, from design through operation, and it treats classical ML and computer vision alongside LLMs. It is lighter than AI Engineering on LLM evaluation and does not cover agents or front-end work. What the book covers lists every chapter.

Which book should you read first?

Read one book properly and build something alongside it before starting the next. The roadmap from software engineer to AI engineer puts reading and projects in sequence.

Frequently asked questions

Sources

1. Chip Huyen: books
2. Chip Huyen: AI Engineering book repository
3. Manning: Build a Large Language Model (From Scratch)
4. Hands-On Large Language Models: official repository
5. Packt: LLM Engineer's Handbook official repository
6. Aurélien Géron: Hands-On Machine Learning with Scikit-Learn and PyTorch
7. Martin Kleppmann: Designing Data-Intensive Applications, second edition
8. Google Research: Reliable Machine Learning