AI Engineering: Building Applications with Foundation Models

Recent breakthroughs in AI have transformed the landscape of application development, making powerful foundation models accessible to developers with varying levels of expertise. In this comprehensive guide, Chip Huyen explores the emerging field of AI engineering—the specialized discipline focused on building robust applications using readily available foundation models rather than creating models from scratch. The book provides a clear distinction between traditional machine learning engineering and this new paradigm, helping readers understand when and how to leverage existing models effectively.

Huyen takes readers through the complete AI application development lifecycle, starting with fundamental concepts and progressing to advanced implementation strategies. She addresses critical challenges such as model evaluation, where traditional metrics fall short with open-ended AI systems, and introduces practical approaches like the rapidly evolving AI-as-a-judge methodology. The book also covers essential adaptation techniques including prompt engineering, retrieval-augmented generation (RAG), fine-tuning, and agent-based systems, explaining not just how they work but when to apply each approach for optimal results.

One of the book’s most valuable contributions is its focus on real-world deployment considerations that often get overlooked in theoretical discussions. Huyen examines the practical bottlenecks of latency and cost when serving foundation models at scale, offering concrete strategies to overcome these challenges. She provides frameworks for selecting appropriate models, datasets, and evaluation benchmarks based on specific application requirements, helping developers make informed decisions in an increasingly crowded marketplace of AI tools and services.

Drawing from her extensive experience at NVIDIA, Snorkel AI, and as a founder of an AI infrastructure startup, Huyen shares insights that bridge the gap between academic research and production implementation. The book includes practical guidance for navigating the complex ecosystem of foundation models, with attention to security considerations, handling hallucinations, and implementing effective guardrails. Her teaching background at Stanford shines through in the clear, structured presentation that makes complex concepts accessible without oversimplification.

This work builds upon and complements Huyen’s previous bestseller, “Designing Machine Learning Systems,” but stands on its own as an essential resource for the new era of AI development. Whether you’re an experienced ML engineer transitioning to foundation models, a developer looking to integrate AI capabilities into existing applications, or a technical manager overseeing AI initiatives, this book provides the practical knowledge needed to move beyond prototype stage to production-ready AI solutions that deliver real business value.