This is a small book on the broader topic of reinforcement learning (RL), written by a practitioner for practitioners. It is very practically oriented, but with limited theoretical background. The book narrative is built around OpenAI Gym, a popular toolkit for the development of RL algorithms. The text gives plenty of technical details related to the libraries, environments, and so on, along with examples in Python.

The first chapter provides a general and granular formulation of RL, but is far from providing deep or comprehensive descriptions of this machine learning technique. Chapter 2 is an overview of common RL algorithms. Here, the author very concisely describes each of the main RL algorithms without diving into long discussions or arguments. However, despite the focus on RL, readers can still get brief introductions to more common optimization algorithms such as gradient descent. A plus for this chapter is its practical discussion of cloud resources. The author gives useful guidelines for docketing experiments in Amazon Web Services (AWS) and/or Google Cloud. As such, this part of the book can be used not only as a quick reference, but also as a short practical guide. However, the interested reader should look elsewhere for more theoretical details.

The book is not a textbook; it is oriented toward enthusiasts and data scientists. The most interesting and most valuable chapter (3) discusses several variants of a popular Q-learning algorithm. The book has a large appendix with source code, which could be very valuable for beginners who want practice. Overall, this is a good book for practitioners who may be interested in developing code for RL.