MLOpsIntermediateOne To One
MLOps: From Notebook to Production
Take a model you trained and ship it as a tested, versioned and monitored service.
About this programme
Fifteen sessions, one project of your own. You start with a model in a notebook and finish with a pipeline that retrains it, a service that serves it, and monitoring that tells you when it drifts. Every session ends with something running on your machine, and you explain back what you built before we move on. This follows the MLOps course I teach to third year B.Tech students at Bennett University, cut down to what you need in practice.
What you will be able to do
- Turn a notebook into a package that runs the same on any machine
- Track experiments, data and model versions with MLflow and DVC
- Serve a model behind a tested API in a Docker container
- Build a pipeline that tests and deploys every change automatically
- Monitor a live model, detect drift and retrain it
Curriculum
5 modules · 15 lessons
Start from the notebook you have and make it something another person can run.
- Turning a notebook into a Python package1h
- Reproducible environments and pinned dependencies1h
- Configuration, logging and a clean project layout1h
Before you start
- Python, including functions, classes and virtual environments
- Training a basic model with scikit learn
- The command line and Git basics