Volume 11 — MLOps¶
Models in production
Experiment tracking, model registries, serving, monitoring, drift detection and retraining — the operational half of machine learning that decides whether any of it survives contact with users.
Outlined, not yet written
This volume is fully outlined but the lessons are not yet authored. The chapter list below is the plan of record, not a promise that content exists.
What you will be able to do¶
- Make an experiment reproducible by someone else
- Deploy a model behind an API with monitoring
- Detect drift before your users do
- Roll back a bad model safely
Chapters¶
- Reproducibility and experiment tracking
- Data and model versioning
- Model registries
- Serving: batch, online and streaming
- Feature stores
- Monitoring and drift detection
- Retraining strategies
- Shadow deployment and canaries
- Incident response for ML systems
- Cost management