Generative AI Systems: Building Generative AI Across Every Modality: from text and images to video, audio, code, 3D, science, and the systems and ethi (Paperback) (ISBN-13: 9798182112765)

Generative AI Systems: The Complete Guide
Building Generative AI Across Every Modality

Behind every generative AI system - the chatbots and image generators, the video, music, and voice models, the coding assistants and autonomous agents - lies the same small set of ideas. Learn them once, and the entire field falls into place.

This is the rare book that covers all of generative AI, not one corner of it: eighty-three chapters across eighteen parts - from first principles to production systems, the industry that builds them, and the research frontier.

Its foundation is a single unifying insight. Almost all generative AI runs on two engines - the autoregressive transformer and the diffusion model - wrapped in one repeatable recipe: choose a representation, apply an engine, condition it for control, and evaluate against the right bar. Master that pattern and every new model, however novel it looks, becomes a variation on a theme you already understand - an approach built to outlast the tools and product names that change every few months.

Inside, you'll build real understanding across the whole field: Foundations - Text & Language (transformers, fine-tuning, prompting, RAG, agents) - Images, Video & 3D (diffusion, latent space, motion, meshes, CAD) - Audio & Music - Code - Data & Knowledge - Interactive & Games - Business, Science & MLOps - Ethics, Safety & the Future - Frameworks & Open Source (LangChain, LangGraph, the library stack) - The Generative AI Industry (companies, business models, technologies) - Training at Scale (the cloud and Kubernetes) - a step-by-step Pseudocode Compendium for every branch - and Frontier Research (mixture-of-experts, reasoning models, flow matching, agents, open problems).

Every chapter is richly illustrated with diagrams, tables, and worked examples, building into one coherent, continuously numbered progression.

Written for developers, engineers, founders, and ML practitioners who want to build generative AI - not just use it. Whether you're shipping your first model or architecting systems at scale, you'll gain both the durable mental models and the engineering discipline to do it well.

The models will keep changing. These principles won't. Start building.