Machine Learning in Production: From Models to Products

by Christian Kästner, Carnegie Mellon University

What does it take to build software products with machine learning, not just models and demos? We assume that you can train a model or build prompts to make predictions, but what does it take to turn the model into a product and actually deploy it, have confidence in its quality, and successfully operate and maintain it at scale? This book explores designing, building, testing, deploying, and operating software products with machine-learned models. It covers the entire lifecycle from a prototype ML model to an entire system deployed in production. Covers also responsible AI (safety, security, fairness, explainability) and MLOps.

To be released in book form by MIT Press.

The book corresponds to the CMU course 17-645 Machine Learning in Production (crosslisted as 11-695 AI Engineering) with publicly available slides and assignments. See also our annotated bibliography on the topic.