MLFlow is an open-source Python framework for experiment tracking. It allows data science teams to store results, artifacts (machine learning models, figures, tables), and metadata in a principled way when executing data pipelines.

The MLFlow plugin for Apache Hamilton includes two sets of features:

• Save and load machine learning models with the MLFlowModelSaver and MLFlowModelLoader materializers
• Automatically track data pipeline results in MLFlow with the MLFlowTracker.

This pairs nicely with the HamiltonTracker and the Apache Hamilton UI which gives you a way to explore your pipeline code, attributes of the artifacts produced, and execution observability.

We're working on better linking Apache Hamilton "projects" with MLFlow "experiments" and runs from both projects.

1. Create a virtual environment and activate it

   python -m venv venv && . venv/bin/active

2. Install requirements for the Apache Hamilton code

   pip install -r requirements.txt

3. Explore the notebook tutorial.ipynb

4. Launch the MLFlow user interface to explore results

   mlflow ui

• Learn the basics of Apache Hamilton via the Concepts/ documentation section
• Visit tryhamilton.dev for an interactive tutorial in your browser
• Visit the DAGWorks blog for more detailed guides