MLOps & Production Systems
Ship models that keep working
Associate Professor · Systems & Optimisation
About this course
Reproducible training, feature pipelines, model registries, serving under latency budgets, monitoring, drift detection and incident response. Taught with the tools teams actually use and the war stories that explain why they use them.
What you will be able to do
- Design reproducible training pipelines with lineage.
- Serve models with SLOs, canaries and rollbacks.
- Monitor drift and run incident response for ML.
Curriculum
5 lessons · 1h 18m
- 01
Reproducibility
- Pipelines, lineage and registriesPreview22 min
- Feature stores, honestly14 min
- 02
Serving and monitoring
- Serving under a latency budget20 min
- Drift, monitoring and incidents16 min
- Module quiz · Operations6 min