Chapter 12: Machine Learning Automated Workflows

For machine learning (ML) models that are deployed to production environments, it's important to establish a consistent and repeatable process to retrain, deploy, and operate these models. This becomes increasingly important as you scale the number of ML models running in production. The machine learning development lifecycle (ML Lifecycle) brings with it some unique challenges in operationalizing ML workflows. This will be discussed in this chapter. We will also discuss common patterns to not only automate your ML workflows, but also implement continuous integration (CI) and continuous delivery/deployment (CD) practices for your ML pipelines.

Although we will cover various options for automating your ML workflows and building CI/CD pipelines for ML in this chapter, we will focus particularly on detailed implementation patterns using Amazon SageMaker Pipelines and Amazon SageMaker projects. SageMaker Pipelines is purpose-built...