Название: MLOps with Databricks: Machine Learning and GenAI Applications End to End
Автор: Maria Vechtomova
Издательство: O’Reilly Media, Inc.
Год: 2026
Страниц: 382
Язык: английский
Формат: epub
Размер: 14.1 MB

MLOps engineers have to deal with a glut of tools and SaaS applications, not to mention technical debt clogging the system. Such complexity requires a comprehensive approach. The Databricks platform provides all the critical components for end-to-end MLOps and LLMOps in one place. This exhaustive book shows you how to use Databricks to build and manage a robust ML system that delivers on your business's needs.

Maria Vechtomova guides you through MLOps principles and explains how Databricks handles the Machine Learning lifecycle holistically, from data preparation to model deployment and monitoring, and enables data engineers, data scientists, and MLOps engineers to collaborate seamlessly. To put all the pieces together, you'll navigate two ML projects: a real-time ML application and an LLM-based system that highlights LLM-specific Databricks features.

Understand the Databricks components for MLOps and LLMOps
Unpack ML Model Serving architectures
Track your machine learning experiments and register your models
Build an ML application that uses Feature and Model Serving, and Model Serving with automatic feature lookup
Deploy a real-time ML application and an LLM-based application
Monitor your AI applications on Databricks
Understand how MLOps principles fit into AI governance

Who and What This Book Is For:
This is a technical book. That means you should have some working experience in data science, data, ML, or AI engineering. All the code covered in the book is written in Python, which is commonly used for developing AI applications. It makes it easier to go through the book if you have already worked with Databricks before, but it’s not a prerequisite.

This is also a book with a lot of code. To benefit from it the most, you should dive into the code and try it out step by step. On the other hand, I do not want you to execute the code blindly. That’s why this book focuses on the MLOps principles first and explains how using different components of Databricks in the right way helps support these principles. This book also covers architectures and offers different considerations for why a certain architecture would be a good choice for a particular use case. Data and AI architects will also benefit from the book even if they don’t try out the code.

Writing a technical book covering different platform features of Databricks is not easy. Based on my experience, this is one of the fastest-evolving products. I realize that parts of the book may become outdated, and I hope to minimize this risk by taking a “principles first” approach. This means that although some of the code, tools, and technical details may become out of date, the theoretical foundations and best practices presented should not. I believe that the book will help a lot of data and AI professionals improve their ML workflows by making them more robust and reliable.

I have written this book as the handbook I wish I’d had when I was first getting started with Databricks. One of the pain points I’ve found is that while there are plenty of examples of how to use different components of Databricks, there are very few examples that bring a cohesive view of how all these components actually work together. I hope this book will provide those examples for you, save you a lot of time, and help you on your MLOps journey.

"This book goes beyond theories. If you want to learn the battle-tested best practices of MLOps and LLMOps, it will walk you through how to implement them with popular Python libraries and deploy them to Databricks. It’s an essential guide in the agentic era." - Jason Yip, Databricks MVP and director of data and AI, Tredence

Ссылки удалены по требованию правообладателя

ОТСУТСТВУЕТ ССЫЛКА/ НЕ РАБОЧАЯ ССЫЛКА ЕСТЬ РЕШЕНИЕ, ПИШЕМ СЮДА!