Documentation

Build on Polnor.

Polnor is a sovereign European Lakehouse platform. The control plane runs in France; everything that touches your data, compute, storage, warehouses, runs in your cloud. This is everything you need to ship on it.

New here? Start with the Quickstart to run your first query in three commands, then read Core concepts to understand workspaces, the control plane and the data plane.

Guides

Connect your cloud Bring OVH or Scaleway credentials and a bucket. Open→ Lakehouse & catalog Register Iceberg tables in the Lakekeeper catalog. Open→ SQL warehouses Query with DuckDB, write with Spark, on demand. Open→ Compute GPU and CPU instances on your own cloud. Open→ Notebooks Git-backed notebooks with live kernels. Open→ Jobs & pipelines Multi-task DAGs with retries and timeouts. Open→ Models & serving MLflow tracking, registry and endpoints. Open→ Governance Audit, lineage, quotas, isolation. Open→

Reference

API reference, the REST API: auth, compute, SQL, jobs, models, lineage.
SDK & CLI, the Python SDK (polnor.mlflow), the Go SDK, and the polnor CLI.

How Polnor is built

A hybrid SaaS, but, unlike the others, the data plane is entirely yours.

Control plane: the API, scheduler, Iceberg catalog, audit and billing, hosted by us on OVHcloud in France.
Data plane: an agent, SQL warehouses and storage that run entirely inside your own OVH or Scaleway account, connected to the control plane over an outbound WebSocket.

Your data never transits our infrastructure. We orchestrate; you own the bytes. See Core concepts for the full picture.