Coming 2027
AI Engineering Cookbook
by Bruno Gonçalves · Pearson
ISBN 978-0-13-618457-7
Eighteen end-to-end recipes for building AI systems that hold up in production. Each chapter takes one real problem — triaging support tickets, answering questions over internal docs, reviewing code, migrating a codebase — and builds a working solution, from a single structured LLM call all the way to multi-agent and self-hosted systems.
18 recipes
Each one a complete, realistic project rather than a toy snippet, so you see how the pieces fit together.
Simple to sophisticated
Eight parts that climb from single-shot calls through RAG, pipelines, and evals to tool-using and multi-agent systems.
Measured, not guessed
Evaluation gets its own part: eval harnesses, statistical comparison, and hallucination detection for RAG outputs.
Inside the book
Table of contents
- Part 1
Single-shot LLM calls — Structured outputs
- 1 Customer Support Ticket Triage and Routing
- 2 Financial Filing Summarizer
- Part 2
RAG and retrieval-augmented workflows
- 3 Internal Documentation Q&A Bot
- 4 Scientific Paper Literature Review Assistant
- Part 3
LLM pipelines — Multi-step orchestration
- 5 Automated Data Cleaning and Schema Inference
- 6 Code Review Assistant with Static Analysis Fusion
- Part 4
LLM-as-judge and evaluation systems
- 7 Eval Harness and Statistical Comparison
- 8 Hallucination Detector for RAG Outputs
- Part 5
Tool-using agents — Single agent, multiple tools
- 9 Spreadsheet Analysis Agent
- 10 Data Science Agent for Kaggle-Style Problems
- 11 Bug Reproduction Agent (SWE-bench)
- Part 6
Multi-agent and long-horizon systems
- 12 Multi-Agent Research Team
- 13 Automated Competitive Intelligence Monitor
- 14 Codebase Migration Agent
- Part 7
Self-hosting and fine-tuning
- 15 Self-hosting an Open-weights LLM with vLLM
- 16 Fine-tuning and Distillation
- 17 Local RAG Stack with Open Embeddings and Open LLM
- Part 8
Capstone
- 18 Knowledge-Base Construction from Unstructured Corpora
Be the first to know
Release dates, early chapters, and pre-order links go out first in the Data Science Briefing, Bruno’s weekly newsletter on LLMs, machine learning, and data science.
Want these techniques in your team before the book ships? See training and consulting.