Hands-On AI Engineering is a practical, code-first guide to building production-grade LLM systems, written by four practicing AI engineers. It focuses on what AI teams deal with every day: performance limits, reliability, evaluation, and cost control.
You'll learn how to design, build, and operate LLM systems that run efficiently, scale responsibly, and hold up under real users - without relying on expensive cloud credits or black-box APIs.
What this book coversTraining and fine-tuning neural networks with PyTorchFine-tuning transformers using LoRA and QLoRA on consumer hardwareBuilding robust RAG pipelines: chunking strategies, hybrid retrieval, ranking, and faithfulness checksDeploying models with FastAPIEvaluating systems properly: rubrics, LLM-as-a-judge, golden datasets, regression testing, benchmarkingMonitoring, failure handling, and cost-performance trade-offsDocumenting architectures and decisions so teams can trust and extend your work Performance add-ons (last chapter)A free companion GitHub repository, carefully sequenced projects you can follow along with and build yourself.
Project 1 - Simple Companion Chat: Basic chatbot built around a single document.Project 2 - Personal Knowledge Q&A: Ask questions over your own files with grounded answers.Project 3 - Checked Q&A System: Compare AI answers against expected results.Project 4 - Conversational Agent: Multi-turn chat with memory and simple tools.Project 5 - Document Summarizer: Controlled summaries with basic quality checks.Project 6 - Chapter Explorer: Turn text into outlines and short quizzes.This book gives you the engineering mindset needed to move from experiments to dependable systems.
The projects are designed to reflect real-world workflows which you can discuss confidently in interviews and use to stand out as an AI engineer.
Use wisely.