In Manning Early Access Program — full release expected January 2027

Most reinforcement learning material stops at games and simulations. This book is about the other part: framing a real business problem as a sequential decision problem, building an environment that respects its constraints, and getting a policy into production that people will actually trust.

What it covers

The book works through reinforcement learning as a practical tool for optimization problems — warehouse order picking, dynamic pricing, bin packing, resource allocation, ad optimization, fuel scheduling, vehicle routing, treatment optimization, data-centre cooling — and then extends the same machinery to fine-tuning large language models.

You’ll learn how to:

  • Frame real business problems as Markov decision processes
  • Design custom environments with domain constraints and reward engineering
  • Apply classical and tabular methods — dynamic programming, bandits, Q-learning, SARSA and Monte Carlo tree search
  • Move to deep RL with DQN and policy gradient methods for high-dimensional problems
  • Fine-tune language models with PPO, GRPO and reinforcement learning from verifiable rewards

It assumes you know your business domain and can program at an intermediate level. It does not assume graduate mathematics.

Contents

Part 1 — Fundamentals: building a reinforcement learning toolkit

  Chapter
1 Real-world decision making with reinforcement learning
2 Markov decision process: turning problems into solvable models
3 Design custom environments for reinforcement learning algorithms

Part 2 — Reinforcement learning for business optimization

  Chapter
4 Perfect knowledge, optimal policy: dynamic programming
5 Contextual bandit: optimizing stochastic one-step decisions
6 Tabular reinforcement learning
7 Monte Carlo tree search: searching with RL principles

Part 3 — Deep reinforcement learning for business optimization

  Chapter
8 Deep Q-networks for high-dimensional data
9 The calculus of decisions: policy gradient methods

Part 4 — Reinforcement learning for LLM fine-tuning

  Chapter
10 Fine-tuning large language models with PPO
11 Reinforcement learning with verifiable feedback using GRPO
12 Building RL systems: an LLM-guided agent for portfolio management

Code

Every chapter from 2 onwards has runnable code in the companion repository, organized by chapter, along with a notebook covering the Python you need for the business applications in the book.

Details

   
Publisher Manning Publications
ISBN 9781633434844
Status Manning Early Access Program, full release expected January 2027
Author Hadi Aghazadeh

Chapters are released as they’re finished through Manning’s early access program, so buying now gets you the current draft plus everything added between now and release.