Summary

This chapter provided an overview of PostgresML, a unique MLOps platform that allows training and calling models from SQL queries on top of an existing PostgreSQL database.

We discussed the platform’s advantages in simplifying an ML-enabled landscape and reducing overhead and network latency in a service-oriented architecture. An overview of the core features and the API was provided.

This chapter concluded with a practical example of leveraging PostgresML for a classification problem, illustrating how to train a LightGBM model, perform hyperparameter optimization, deploy it, and leverage it for predictions in a handful of SQL queries.

In the next chapter, we will look at distributed and GPU-based learning with LightGBM.