Case Study: Photomath speeds up ML experimentation with ClearML

How Photomath Uses ClearML to Iterate ML Experiments Quickly

Photomath, the world’s most-used math education app, needed a better way to manage ML experiments for its computer vision models. Before adopting ClearML, the team tracked metadata in Google Sheets, struggled with lost or conflicting experiments, and spent too much time manually comparing runs and dealing with inconsistent data.

Using ClearML’s open source MLOps platform, Photomath was able to run experiments remotely, track metrics, compare experiments, manage artifacts and datasets, and share results more easily across the team. ClearML helped boost experimentation by 1200% from two runs per day to dozens, while speeding up graphing and results analysis by 1500%, from about 15 minutes to just seconds.

Photomath

Marijan Smetko

Junior AI Engineer