Introducing Model Serving Patterns
Serving a machine learning model is one of the most complex steps in the machine learning life cycle. In Chapter 1, we saw why serving machine learning models is challenging. Serving machine learning models involves two groups in two different domains: the ML developer develops the model and the software developer serves the model. So, we need to agree upon a common language so that we can be sure how our model will be deployed to solve a particular kind of problem. Patterns in design help software architects systematically solve complicated software engineering problems. Similarly, as we learn about design patterns in model serving, the complicated process of model serving will eventually become a piece of cake. This chapter will build on the ideas of some already used patterns for ML serving. We collect the patterns followed by developers and organize and classify those patterns. This chapter will discuss the following topics:
- Design patterns...
