Introduction
In the previous chapters, you have learned ways to set up a data storage environment for AI. In this chapter, we will explore the final step: taking machine learning models into production, so that they can be used in live business applications. There are several methods for productionizing models, and we will elaborate on a few common ones.
Data scientists are trained to wrangle data, pick a machine learning algorithm, do feature engineering, and optimize the models they create. But even an excellent model has no value if it only runs in a machine learning environment or on the laptop of the data scientist; it has to be deployed in a production application. Furthermore, models have to be regularly updated to reflect the latest feedback from customers. Ideally, a model is continuously and automatically refreshed in a feedback loop; we call that reinforcement learning. An example of a system that uses reinforcement learning is a recommendation engine on a video...