📄️ Complete MLOps
Take a churn model from a Jupyter notebook to production: tracking, versioning, CI/CD, containers, drift monitoring, automated retraining and rollback.
📄️ 1. What MLOps solves
Module 1 of the Complete MLOps premium course: why the discipline exists, what makes ML systems different from regular software, and what a starting notebook lacks.
📄️ 2. Reproducibility
Module 2 of the Complete MLOps premium course: seed every source of randomness, pin every dependency, and understand why GPUs stay non-deterministic by default.
📄️ 3. MLflow tracking
Module 3 of the Complete MLOps premium course: log parameters, metrics and artifacts of every run, compare runs, use autologging, and share a tracking server.
📄️ 4. Data and model versioning
Module 4 of the Complete MLOps premium course: DVC with a remote store, linking commit and data and model, lineage, and when a versioned database is enough.
📄️ 5. Model registry
Module 5 of the Complete MLOps premium course: the MLflow registry, aliases and stages, approval gates, rollback, and the metadata a registered model must carry.
📄️ 6. Containerizing a model
Module 6 of the Complete MLOps premium course: build a minimal service image, pin dependencies, load the model at startup, and manage image size and security.
📄️ 7. CI/CD for ML
Module 7 of the Complete MLOps premium course: data tests, model tests with thresholds, image build, deployment with a manual gate, all in GitHub Actions.
📄️ 8. Batch, online, streaming
Module 8 of the Complete MLOps premium course: pick a serving mode by latency requirements, understand feature consistency, and see when batch beats online.
📄️ 9. Monitoring and drift
Module 9 of the Complete MLOps premium course: statistical tests per feature, target drift, useful alerts, Evidently reports, latency and application errors.
📄️ 10. Retraining and rollback
Module 10 of the Complete MLOps premium course: triggers for retraining, validation before promotion, canary deployments, one-command rollback, cost and governance.
📄️ Recap and exam
Complete recap of the Complete MLOps premium course: reproducibility, tracking, versioning, registry, container, CI/CD, serving, monitoring, retraining, and the 40-question exam.