Book Image

Engineering MLOps

By : Emmanuel Raj
Book Image

Engineering MLOps

By: Emmanuel Raj

Overview of this book

Engineering MLps presents comprehensive insights into MLOps coupled with real-world examples in Azure to help you to write programs, train robust and scalable ML models, and build ML pipelines to train and deploy models securely in production. The book begins by familiarizing you with the MLOps workflow so you can start writing programs to train ML models. Then you’ll then move on to explore options for serializing and packaging ML models post-training to deploy them to facilitate machine learning inference, model interoperability, and end-to-end model traceability. You’ll learn how to build ML pipelines, continuous integration and continuous delivery (CI/CD) pipelines, and monitor pipelines to systematically build, deploy, monitor, and govern ML solutions for businesses and industries. Finally, you’ll apply the knowledge you’ve gained to build real-world projects. By the end of this ML book, you'll have a 360-degree view of MLOps and be ready to implement MLOps in your organization.
Table of Contents (18 chapters)
1
Section 1: Framework for Building Machine Learning Models
7
Section 2: Deploying Machine Learning Models at Scale
13
Section 3: Monitoring Machine Learning Models in Production

Maintaining the CI/CD pipeline

As you may recall, in Chapter 10, Essentials of Production Release, we mentioned that a model is not the product; the pipeline is the product. Hence, after setting up automated or semi-automated CI/CD pipelines, it is critical to monitor the performance of our pipeline. We can do that by inspecting the releases in Azure DevOps, as shown in the following screenshot:

Figure 13.18 – Maintaining CI/CD pipeline releases

The goal of an inspection is to keep the CI/CD pipeline in a healthy and robust state. Here are some guidelines for keeping the CI/CD pipeline healthy and robust:

  • If a build is broken, a fix it asap policy from the team should be implemented.
  • Integrate automated acceptance tests.
  • Require pull requests.
  • Peer code review each story or feature.
  • Audit system logs and events periodically (recommended).
  • Regularly report metrics visibly to all the team members (for example, slackbot or...