1 · Foundations
What MLOps is, the ML lifecycle, and the reproducible project structure every deployment rests on.
●Track 7 · live
MLOps & Deployment
Ship a model to the cloud — Docker, CI/CD, MLflow, AWS.
Docker basics through to deployment on AWS Beanstalk, EC2 + ECR, and Azure containers. Experiment tracking with MLflow + DagsHub, data versioning with DVC. Read-along code paired with animated diagrams of Docker layers, ECR push flow, and CI pipelines.
18 / 18 lessons8 chapters
2 · Docker for ML
Package a model app into a portable image — Dockerfiles, layer caching, multi-stage builds, and docker-compose.
3 · Serving the model
Expose predictions over HTTP with a FastAPI endpoint, then containerize it with health checks.
4 · Experiment tracking
Stop guessing which run was best — log params, metrics, and artifacts with MLflow and promote models through a registry.
5 · Data & model versioning
Version large data and models alongside code with DVC, and make the whole pipeline reproducible.
6 · CI/CD
Automate testing and deployment with GitHub Actions — run checks on every PR, then build, push, and ship on merge.
7 · Cloud deployment
Ship the container to the cloud — push to a registry (ECR), then deploy on AWS (Beanstalk, EC2, SageMaker) or Azure managed endpoints.
8 · Monitoring & capstone
Watch a live model for data and concept drift, then tie the whole track together into one end-to-end path from training to a monitored, self-retraining production model.