What do you think?
Rate this book
Machine Learning Platform Engineering: Build an internal developer platform for ML and AI systems
Get your machine learning models out of the lab and into production!
Delivering a successful machine learning project is hard. Build a Machine Learning Platform (From Scratch) makes it easier. In it, you’ll design a reliable ML system from the ground up, incorporating MLOps and DevOps along with a stack of proven infrastructure tools including Kubeflow, MLFlow, BentoML, Evidently, and Feast.
In Build a Machine Learning Platform (From Scratch) you’ll learn how
• Set up an MLOps platform
• Deploy machine learning models to production
• Build end-to-end data pipelines
• Effective monitoring and explainability
A properly designed machine learning system streamlines data workflows, improves collaboration between data and operations teams, and provides much-needed structure for both training and deployment. In Build a Machine Learning Platform (From Scratch) you’ll learn how to design and implement a machine learning system from the ground up. You’ll appreciate this instantly-useful introduction to achieving the full benefits of automated ML infrastructure.
About the book
Build a Machine Learning Platform (From Scratch) teaches you to set up and run a production-quality machine learning system using open source tools. Chapter-by-chapter, you’ll assemble a delivery pipeline for an image classifier and a recommendation system, learning best practices as you go. Whether you're working with traditional models or tackling the creation of a cutting-edge transformer like the one detailed in Sebastian Raschka’s Build a Large Language Model (From Scratch), this book provides the crucial MLOps framework to get it into production. You’ll get hands-on experience with the most important parts of the machine learning workflow, including orchestrating pipelines; model training, inference, and serving; and monitoring and explainability. Soon, you’ll be deploying models that are fast to production and easy to maintain and scale.
About the reader
For data scientists or software engineers who know how to program in Python.
About the author
Benjamin Tan is a product manager and principal engineer for sata Science at DKatalis where he leads a team of talented machine learning engineers, data scientists, and data engineers. He is also the author of The Little Elixir and OTP Guidebook and Building an ML Pipeline with Kubeflow (liveProject) from Manning, and Mastering Ruby Closures.
Shanoop Padmanabhan is a software engineering manager at Continental Automotive, where he leads a team of software engineers focusing on machine learning based perception for autonomous vehicles.
Varun Mallya is a machine learning engineer working at DKatalis where he is responsible for the setup and maintenance of the Bank’s machine learning platform.
Delivering a successful machine learning project is hard. Build a Machine Learning Platform (From Scratch) makes it easier. In it, you’ll design a reliable ML system from the ground up, incorporating MLOps and DevOps along with a stack of proven infrastructure tools including Kubeflow, MLFlow, BentoML, Evidently, and Feast.
In Build a Machine Learning Platform (From Scratch) you’ll learn how
• Set up an MLOps platform
• Deploy machine learning models to production
• Build end-to-end data pipelines
• Effective monitoring and explainability
A properly designed machine learning system streamlines data workflows, improves collaboration between data and operations teams, and provides much-needed structure for both training and deployment. In Build a Machine Learning Platform (From Scratch) you’ll learn how to design and implement a machine learning system from the ground up. You’ll appreciate this instantly-useful introduction to achieving the full benefits of automated ML infrastructure.
About the book
Build a Machine Learning Platform (From Scratch) teaches you to set up and run a production-quality machine learning system using open source tools. Chapter-by-chapter, you’ll assemble a delivery pipeline for an image classifier and a recommendation system, learning best practices as you go. Whether you're working with traditional models or tackling the creation of a cutting-edge transformer like the one detailed in Sebastian Raschka’s Build a Large Language Model (From Scratch), this book provides the crucial MLOps framework to get it into production. You’ll get hands-on experience with the most important parts of the machine learning workflow, including orchestrating pipelines; model training, inference, and serving; and monitoring and explainability. Soon, you’ll be deploying models that are fast to production and easy to maintain and scale.
About the reader
For data scientists or software engineers who know how to program in Python.
About the author
Benjamin Tan is a product manager and principal engineer for sata Science at DKatalis where he leads a team of talented machine learning engineers, data scientists, and data engineers. He is also the author of The Little Elixir and OTP Guidebook and Building an ML Pipeline with Kubeflow (liveProject) from Manning, and Mastering Ruby Closures.
Shanoop Padmanabhan is a software engineering manager at Continental Automotive, where he leads a team of software engineers focusing on machine learning based perception for autonomous vehicles.
Varun Mallya is a machine learning engineer working at DKatalis where he is responsible for the setup and maintenance of the Bank’s machine learning platform.
923 pages, Kindle Edition
Published March 17, 2026
6 people are currently reading
6 people want to read
About the author
Benjamin Tan Wei Hao
3 books11 followersBenjamin is a Software Engineer at Pivotal Labs, Singapore.
He is the author of The Little Elixir and OTP Guidebook (Manning Publications) and The Ruby Closures Book (The Pragmatic Bookshelf).
Deathly afraid of being irrelevant, is always trying to catch up on his ever-growing reading list.
He is the author of The Little Elixir and OTP Guidebook (Manning Publications) and The Ruby Closures Book (The Pragmatic Bookshelf).
Deathly afraid of being irrelevant, is always trying to catch up on his ever-growing reading list.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 8 of 8 reviews
April 8, 2026
This book has lots of useful information, and details on Kubernetes, docker and monitoring with Grafana etc. I don't know these inside out, so had to take it slowly, hence 4 stars not 5.
This is a fast moving topic/area but this is a useful book. Seeing ways to version control your ML models is very important.
.
This is a fast moving topic/area but this is a useful book. Seeing ways to version control your ML models is very important.
.
April 21, 2026
The field of Machine Learning Platform Engineering is far from trivial, but after finishing the first six chapters, I can say it is an incredibly engaging read. As someone with limited experience in the field, I found the topic challenging yet rewarding. The author does an excellent job of capturing the reader’s attention despite the inherent complexity of the subject matter. What sets this book apart is its hands-on approach. The author guides you through building actual models which, while simple in architecture, serve as the perfect foundation for understanding the broader infrastructure. This is a demanding read. To successfully implement the examples, you will need a solid foundation in Python. Being able to both read and write code comfortably is essential to keeping up with the technical requirements of the platform implementation. I recommend this book to anyone looking to understand the "how" and "why" of ML infrastructure. It is a steep learning curve, but it is effectively designed for those who want to transition from simple model building to professional-grade platform engineering
March 18, 2026
This book isn't going to hold your hand through every step — and honestly, if you go in expecting that, you're gonna have a bad time. But if you're cool with rolling up your sleeves and doing some extra digging, there's a lot to learn here.
The real value is in the architecture. The way the authors break down how an ML platform fits together genuinely clicked for me, and that mental model alone is worth a decent chunk of the cover price. It's the kind of stuff that's hard to piece together from blog posts and Stack Overflow threads.
Yeah, the code samples have some bugs, and the book moves fast enough that you'll occasionally feel like it's assumed you already know the thing it's supposed to be teaching you. Some sections lean a bit too heavily on just clone this repo and run it, which works until it doesn't, and then you're on your own. A bit more explanation of why things are configured the way they are would've gone a long way.
The real value is in the architecture. The way the authors break down how an ML platform fits together genuinely clicked for me, and that mental model alone is worth a decent chunk of the cover price. It's the kind of stuff that's hard to piece together from blog posts and Stack Overflow threads.
Yeah, the code samples have some bugs, and the book moves fast enough that you'll occasionally feel like it's assumed you already know the thing it's supposed to be teaching you. Some sections lean a bit too heavily on just clone this repo and run it, which works until it doesn't, and then you're on your own. A bit more explanation of why things are configured the way they are would've gone a long way.
March 24, 2026
I work around ML platforms, so I pick up MLOps books once in a while. A lot of them repeat the same lifecycle diagrams and stay pretty high level. This one goes a bit more into the engineering side. The examples helped. The object detection pipeline and the movie recommender made it easier to see how the pieces fit together instead of just reading theory. It walks through how things move from experimentation to something that can actually run in production.
I also liked that the book touches on LLM systems later on. Many MLOps books still ignore that part, so it was good to see some coverage there. One thing to know going in: the stack leans heavily toward K8s. If you haven't spent time around that ecosystem, a few sections might slow you down. Not a deal breaker, just something to expect.
Overall I found it useful. If you're more interested in the platform and engineering side of ML systems than the modeling side, there is good material here.
I also liked that the book touches on LLM systems later on. Many MLOps books still ignore that part, so it was good to see some coverage there. One thing to know going in: the stack leans heavily toward K8s. If you haven't spent time around that ecosystem, a few sections might slow you down. Not a deal breaker, just something to expect.
Overall I found it useful. If you're more interested in the platform and engineering side of ML systems than the modeling side, there is good material here.
Review of advance copy received from Publisher
This book was originally titled “Build a Machine Learning Platform (from Scratch).” The authors state that the book will teach you how to set up an ML platform so that you can learn how to deploy ML models. I am only about one-quarter of the way through it, but there are some major flaws.
First, for installing ArgoCD and Kubeflow, the authors provide a custom ArgoCD repository with custom configuration, as well as one for Kubeflow. In the installation appendix, they simply tell you to blindly follow the instructions to deploy ArgoCD and Kubeflow from the repository without explaining how these are set up. As a result, reading this book will not teach you how to set up an ML platform on your own unless you install these tools separately from scratch yourself.
Therefore, it falls far short of what was promised, which is why I rated it 2/5 stars.
First, for installing ArgoCD and Kubeflow, the authors provide a custom ArgoCD repository with custom configuration, as well as one for Kubeflow. In the installation appendix, they simply tell you to blindly follow the instructions to deploy ArgoCD and Kubeflow from the repository without explaining how these are set up. As a result, reading this book will not teach you how to set up an ML platform on your own unless you install these tools separately from scratch yourself.
Therefore, it falls far short of what was promised, which is why I rated it 2/5 stars.
April 11, 2026
This book stands out as one of the most practical and comprehensive resources on MLOps. Unlike titles that stay high-level, it dives into engineering details with hands-on examples like object detection pipelines and movie recommenders, showing how experimentation evolves into production-ready systems. Coverage spans the full lifecycle—data integration, training, serving, and monitoring—with clear writing and well-structured chapters. The inclusion of LLM systems is a welcome addition, though the stack leans heavily on Kubernetes. While deeper exploration of data infrastructure and emerging AI agent workflows would strengthen it further, the book remains highly valuable. For intermediate to advanced ML engineers seeking a single, grounded resource, this is strongly recommended
March 14, 2026
While I agree the included code samples have some bugs, I’m more positive than previous reviewers. Having read this as part of the editorial review team, I’ll be honest: the code needs a significant refresh—a common hurdle in fast-moving tech books. However, the core methodology and the framework the authors provide are very relevant. Don't expect a 'plug-and-play' experience; there is some 'plumbing' required to get things running. But if you’re willing to put in that effort, the architectural knowledge you’ll gain about ML platforms is really valuable.
April 2, 2026
Machine Learning Platform Engineering starts simple, explaining introductory topics and advancing into the harder ones. The progression of the book is smooth and continuous, without hard cuts. I had the pleasure of being one of the reviewers, and I can confidently say, if you want to become an expert in MLOps, you should read this book.
Displaying 1 - 8 of 8 reviews