← Work

2024 · Research

Experiment tracking and model governance with MLflow

The problem

Machine learning models are hard to reproduce when experiments, settings and files are not tracked in one place.

What I built

I explored how MLflow tracks experiments, parameters and stored files so a training run can be repeated and compared later.

The result

A clear way to repeat and compare model training runs.

Main tools

  • MLflow

This is the short version. The full case study is available on request.

Ask for the full case study →