17-444 Production Machine Learning Systems for Software Engineers

9 units, Carnegie Mellon University

This course prepares software engineers to build and operate production-grade systems powered by machine learning. Assuming prior knowledge of software engineering and basic machine learning, it examines how established engineering practices must adapt to probabilistic, data-dependent, and continuously evolving behavior. Students study the full application lifecycle for systems integrating ML models, large language models, and AI agents. Topics include MLOps practices such as pipeline automation, versioning, reproducibility, deployment, monitoring, and drift detection; harness engineering for increasingly autonomous systems, including orchestration, evaluation, guardrails, permissions, and human oversight; and system-level qualities such as safety, security, privacy, fairness, and explainability.

Prerequisites: 17-313

CMU Courses needs JavaScript to show sections, schedules and planning tools.