Preface

The space of large language models (LLMs) is evolving at a ludicrous pace. As we write this, the DeepSeek team has just released a new paper showing how to compress an LLM’s context using computer vision techniques. The sheer level of activity in the field of artificial intelligence is impressive and shows no signs of slowing down.

As authors of this book, we thought, What can we possibly write about DeepSeek that wouldn’t make this book instantly outdated? Our answer is simple: we believe that the future of AI technology is open, and the pioneer of that movement is DeepSeek.

DeepSeek has not stopped surprising the world of artificial intelligence, from releasing incredibly powerful models to publishing breakthrough research focused on LLMs. Given these developments, it was clear to us that we needed to write a book about what was happening.

The goal of this book is to give you - dear reader - the necessary tools to master DeepSeek. We want it to serve as your guide to mastering open-source language models.

In this book, we cover nearly everything you might encounter while working with DeepSeek models - from understanding what sets DeepSeek models apart to how you can use them for practical applications. We also cover how to master prompting DeepSeek models, so you can become an expert at interacting with this family of models. Finally, we dive deep into the practical side of things, with examples of how to design, build, and deploy agentic and non-agentic applications powered by DeepSeek models.

In this book, we start with an introduction to DeepSeek, where we explore its foundations and understand what sets it apart from the rest (Chapters 1 and 2). Once you understand what DeepSeek can do, it’s time to learn how to use it effectively. We have an entire chapter dedicated to the art of prompting reasoning models, where we show you how less is more (Chapter 3).

We then move into practical applications, demonstrating how to use it to produce consultant-grade analysis of complex problems (Chapter 4). From there, we build a complete end-to-end application where you’ll go from a simple API all the way to a fully containerized service running on Amazon Web Services (Chapter 5).

In the final part of this book, we explore the edge of LLM and AI technologies. You’ll learn how to use DeepSeek models as backbones for agentic applications (Chapter 6). We also walk you through a more complex MLOps use case where you’ll use a more powerful DeepSeek model to distill knowledge into a smaller one (Chapter 7). In the last chapter of this book, we cover the deployment of DeepSeek models, with their many trade-offs, so you can choose the best deployment methodology for your use case (Chapter 8).

To conclude this book we have an Epilogue to walk you through the key takeaways. The book also provides an Appendix that guides you through various ways to use DeepSeek.