Expand MLOps lab and ML Theory hyperparameter tuning

[wypa93]

Summary

• Adds a full MLOps lab notebook flow (MLflow tracking, registry, projects, serving, Evidently drift) with Docker-friendly setup, port mapping docs, and pickle-based model serialization for MLflow compatibility.
• Expands the ML Theory notebook with 5-fold cross-validation, grid search, and Bayesian hyperparameter search via scikit-optimize.
• Updates wine_classifier train/predict scripts, Docker image dependencies (mlflow, evidently, scikit-optimize), and gitignore rules for local lab artifacts.

Test plan

• Build Docker image: docker build -t aikurs-container .
• Run container with ports -p 8888:8888 -p 5050:5050 -p 5001:5001 -p 8080:8080 and open Jupyter with fresh token URL
• Run ML Theory notebook through section 5 (CV, grid search, Bayesian search)
• Run MLOps notebook end-to-end; verify MLflow UI at localhost:5050 and model serve via POST to /invocations
• Run mlflow run from labs/05 MLOps/wine_classifier with MLFLOW_TRACKING_URI set

Made with Cursor