Machine Learning Model Serving Patterns and Best Practices

9

Pipeline Pattern Model Serving

In this chapter, we will discuss the pipeline pattern of model serving. In this pattern, we create a pipeline with a number of stages. In each of the stages, some functionalities (such as data collection and feature extraction) are performed. The pipeline is started using a scheduler or it can be started manually. We will discuss how to create a pipeline for serving machine learning (ML) models and demonstrate how the pipeline pattern can be used for this purpose. The high-level topics that will be covered in this chapter are as follows:

• Introducing the pipeline pattern
• Introducing Apache Airflow
• Demonstrating a machine learning pipeline using Apache Airflow
• Advantages and disadvantages of the pipeline pattern ...