AI Engineering places foundation models within the broader work of building an application. Chip Huyen connects model selection, evaluation, context, data, and serving decisions to the quality, cost, and latency of the resulting product.
From Model Output to Application Behavior
The book covers foundation models, evaluation methods and pipelines, prompting, RAG and agents, fine-tuning, dataset engineering, inference optimization, architecture, and user feedback. Evaluation provides a way to compare changes on a defined task rather than judge them from a convincing demonstration alone.
Its practical value lies in distinguishing different interventions. Adding context, adapting model behavior, and improving the surrounding application solve different problems. The evaluation process helps identify which intervention has made a useful difference.
Reading Context
Software engineers looking beyond individual prompts will find a substantial reference, with some chapters requiring more technical background. Building Applications with AI Agents complements it with a narrower examination of tools, memory, and agent coordination.
Get the Book
- O'Reilly: AI Engineering
- Author's resources: AI Engineering companion repository