Graviti

Sorting, versioning and cleaning up the unstructured data behind a machine learning model is the work Graviti is built around.

Graviti is a data platform for AI and ML teams that handles dataset curation, data version control and workflow automation in one spot.

Its aim is to make messy data, the kind that rarely fits into ordinary analytics tools, usable for both analysis and model training at scale.

Versioning is a central piece.

Graviti ties data versioning to MLflow, so teams already tracking experiments there can keep datasets under the same kind of control.

That helps when several people touch the same data and nobody can say which copy a model was trained on.

Collaboration is the other focus.

Graviti uses data visualization so team members can look at the same datasets instead of working in separate silos.

Curation is described as fast and cost-effective, and workflow automation takes repeated steps off your hands.

Graviti suits teams building AI products who spend more time wrangling data than training models.

It is a developer-facing tool, not a general office app.