MLOps with Red Hat OpenShift

A cloud-native approach to machine learning operations

Ross Brigoli , Faisal Masood 지음

Packt(GCO Science) 2024년 01월 31일 출간

작품소개

책소개

Build and manage MLOps pipelines with this practical guide to using Red Hat OpenShift Data Science, unleashing the power of machine learning workflows

▶Book Description
MLOps with OpenShift offers practical insights for implementing MLOps workflows on the dynamic OpenShift platform. As organizations worldwide seek to harness the power of machine learning operations, this book lays the foundation for your MLOps success. Starting with an exploration of key MLOps concepts, including data preparation, model training, and deployment, you’ll prepare to unleash OpenShift capabilities, kicking off with a primer on containers, pods, operators, and more.
With the groundwork in place, you’ll be guided to MLOps workflows, uncovering the applications of popular machine learning frameworks for training and testing models on the platform.
As you advance through the chapters, you’ll focus on the open-source data science and machine learning platform, Red Hat OpenShift Data Science, and its partner components, such as Pachyderm and Intel OpenVino, to understand their role in building and managing data pipelines, as well as deploying and monitoring machine learning models.
Armed with this comprehensive knowledge, you’ll be able to implement MLOps workflows on the OpenShift platform proficiently.

▶ What You Will Learn
⦁ Build a solid foundation in key MLOps concepts and best practices
⦁ Explore MLOps workflows, covering model development and training
⦁ Implement complete MLOps workflows on the Red Hat OpenShift platform
⦁ Build MLOps pipelines for automating model training and deployments
⦁ Discover model serving approaches using Seldon and Intel OpenVino
⦁ Get to grips with operating data science and machine learning workloads in OpenShift

목차

▶ TABLE of CONTENTS
1. Introduction to MLOps and OpenShift
2. Provisioning an MLOps platform in the Cloud
3. Building Machine Learning Models
4. Embedding ML Models into the Applications
5. Deploying ML Models as a Service
6. Operating ML workloads
7. Building a face detector using the Red Hat ML Platform

책 속으로

▶ What this book covers
⦁ Chapter 1, Introduction to MLOps and OpenShift, starts with a brief introduction to MLOps and the basics of Red Hat OpenShift. The chapter then discusses how OpenShift enables machine learning projects and how Red Hat OpenShift Data Science and partner software products comprise a complete MLOPS platform.
⦁ Chapter 2, Provisioning an MLOps Platform in the Cloud, will walk you through provisioning Red Hat OpenShift, Red Hat OpenShift Data Science, and Pachyderm on the AWS cloud. The chapter contains step-by-step instructions on how to provision the base MLOps platform.
⦁ Chapter 3, Building Machine Learning Models with OpenShift, starts with the initial configurations of the platform components to prepare for model building. The chapter walks you through the configuration steps and ends with an introduction to the data science projects, workbenches, and the Jupyter Notebook.
⦁ Chapter 4, Managing a Model Training Workflow, digs deeper into the platform configuration covering OpenShift Pipelines for building model training pipelines and using Pachyderm for data versioning. By the end of the chapter, you will have built an ML model using a training pipeline you created.
⦁ Chapter 5, Deploying ML Models as a Service, introduces the model serving component of the platform. The chapter will walk you through how to enhance further the pipeline to automate the deployment of ML models.
⦁ Chapter 6, Operating ML Workloads, talks about the operational aspects of MLOps. The chapter focuses on logging and monitoring the deployed ML models and briefly discusses strategies for optimizing operational costs.
⦁ Chapter 7, Building a Face Detector Using the Red Hat ML Platform, walks you through the process of building a new AI-enabled application from end to end. The chapter helps you practice the knowledge and skills you gained in the previous chapters. The chapter also introduces Intel OpenVino as another option for model serving. By the end of this chapter, you will have built an AI-enabled web application running on OpenShift and used all of the Red Hat OpenShift Data Science features.

출판사 서평

▶ Preface
MLOps, or Machine Learning Operations, is all about streamlining and harmonizing the intricate dance between developing and deploying machine learning models. It’s like the conductor orchestrating a symphony, ensuring a seamless flow from the creative realm of data science to the robust reality of IT operations.
This book introduces a practical approach to implementing MLOps on the Red Hat OpenShift platform. It starts by presenting key MLOps concepts such as data preparation, model training, and packaging and deployment automation. An overview of OpenShift’s fundamental building blocks—deployments, pods, and operators—is then provided. Once the basics are covered, the book delves into platform provisioning and deepens our exploration of MLOps workflows.
Throughout the book, Red Hat OpenShift Data Science (RHODS), a data science platform designed to run on OpenShift, is utilized. You will experience creating ML projects, notebooks, and training and deployment pipelines using RHODS. The book also covers the use of partner software components that complement the RHODS platform, including Pachyderm and Intel OpenVino.
By the book’s end, you will gain a solid understanding of MLOps concepts, best practices, and the skills needed to implement MLOps workflows with Red Hat OpenShift Data Science on the Red Hat OpenShift platform.

작가정보

저자(글) Ross Brigoli

Ross Brigoli is a consulting architect at Red Hat, the largest open source software company in the world, delivering innovative solutions to its customers. He has professional experience spanning more than two decades, marked by expertise in software engineering and architecture, solution design, data engineering, machine learning, DevOps, and MLOps. Before Red Hat, Ross was an associate director at Credit Agricole Corporate and Investment Bank, leading the architecture of a big data platform. Ross, along with Faisal Masood, co-authored the book Machine Learning on Kubernetes.

저자(글) Faisal Masood

Faisal Masood is a cloud transformation architect at AWS. Faisal’s focus is on assisting customers in refining and executing strategic business goals. His main interests are evolutionary architectures, software development, the ML life cycle, CD, and IaC. Faisal has over two decades of experience in software architecture and development.