MLflow
Track your experiments and version your models.
MLflow tracks your machine learning experiments, compares runs and manages your model registry, in personal or team workspaces.
- Open source souverain · MLflow (Apache-2.0)
- ○ MCP integration in progress
- Access on request (AI Lab)
Illustrative preview, fictional data
Features
Everything you need, nothing that tracks you.
Experiment tracking
Parameters, metrics and artefacts for each run.
Comparison
Compare runs side by side to pick the best one.
Model registry
Versions, aliases and production rollout.
Separate workspaces
Personal or per organisation, with your permissions.
Linked to notebooks
Log your runs from Jupyter.
Artefacts in Europe
Models and files stored on Eurathos object storage.
Powered by Apollon
Ask, and Apollon handles it in MLflow.
The MCP server for MLflow is being integrated: Apollon will soon be able to act on your behalf, with your approval.
- « Compare my last two runs »
- « What is the best model for the churn experiment? »
- « Set the production alias to version 4 »
Use cases
- Industrialise model training.
- Keep track of every experience across a data team.
- Promote a validated model to production.
The European alternative to
Hosted in Europe
On our servers in France, beyond the reach of the Cloud Act.
Secure single account
Eurathos sign-in with two-factor authentication and passkeys.
Open source souverain · MLflow (Apache-2.0)
Self-hosted, updated and secured by Eurathos.
Saved
Data backed up off-site, restore tested.
Works with
Ready to switch to MLflow ?