Case Study: Uber improves ML experiment management with Comet

How Scientists at Uber Use Comet to Manage Machine Learning Experiments

Uber, a mobility-as-a-service provider in ridesharing and delivery, needed a better way to manage offline machine learning experiments at scale. With many teams, markets, and product categories to evaluate, Uber AI struggled to compare results consistently and needed a solution that could complement its in-house experiment management and collaboration tools. Uber turned to Comet for ML experiment tracking.

Comet helped Uber track hyperparameters, metrics, and code changes across experiments, making it easier to organize deep learning work and analyze model performance by market and product segment. The platform’s custom panels, code panels, and parallel coordinate charts also let Uber visualize hyperparameter searches and add project-specific metrics without changing the core platform, improving developer velocity and supporting deeper experimentation workflows.

Uber

Olcay Cirit

Research Scientist