AI as an Infrastructure
A free, web-only book that explains AI systems, from raw compute to a deployed and governed behavior, design-first, for a software engineer who knows distributed systems but is new to machine learning.
The book follows the lifecycle of a capability as one continuous arc, asking at each step how a piece got its shape, what trade-off that shape encodes, and the theory underneath, holding a consistent introductory level from its first chapter to its last.
It spans thirteen parts and 93 chapters covering base model formation, generative and multimodal architectures, post-training and alignment, reasoning and test-time compute, and inference and serving, among other topics, with runnable cells and interactive figures that only work in the web version.
It is published in parallel English and Chinese editions with the same chapters, sources, equations and figures.