Course Wiki

MLflow

The open-source ML lifecycle toolkit MLOps Zoomcamp standardizes on: tracking, model logging, and a registry.

MLflow is the course’s tool for the experiment-tracking module: mlflow.start_run() wraps a training run, autologging or explicit calls capture parameters and metrics, and mlflow.log_artifact stores the trained model in a standard format. A tracking server (local in the course, remote in production) keeps everything browsable.

The module also covers the practical operations: loading a logged model back for scoring, switching the artifact store to S3, and standing up a remote tracking server so runs from training jobs and deployments land in the same place. Deployment later pulls models from MLflow by run id and stage.

Taught in

• MLOps Zoomcamp

  • Module 2: Experiment Tracking & Model Management
    • Getting started with MLflow
    • Experiment tracking with MLflow
    • Model registry
    • MLflow in practice
    • MLflow: benefits, limitations and alternatives
    • Homework
  • Module 3: Orchestration & ML Pipelines
    • Homework
  • Module 4: Model Deployment
    • Web-services: Getting the models from the model registry (MLflow)
    • MLOps Zoomcamp 4.6 - Batch scoring with Mage
  • Module 07: Course Project
    • Course Project ## Related concepts

• Experiment Tracking

• Model Registry

• MLOps Maturity Model

• Model Deployment