API reference

GET /compute?limit=50&cursor=eyJpZCI6...

# → { "data": [ … ], "next_cursor": "eyJpZCI6…" }

Exchange email and password for a short-lived access token.

Request

curl -s https://api.polnor.net/api/v1/auth/login \
  -H "Content-Type: application/json" \
  -d '{"email":"you@acme.eu","password":"••••••"}'

200 OK

{
  "access_token": "eyJhbGciOi…",
  "token_type": "Bearer",
  "expires_in": 3600
}

Issue a fresh access token before the current one expires.

List compute instances in the workspace.

200 OK

{ "data": [
  { "id": "cmp_8f21", "name": "training-a100",
    "flavor": "gpu.a100", "status": "running",
    "region": "GRA9", "auto_stop": "2h" }
] }

Create an instance. It is provisioned on the cloud provider registered for the workspace.

Body

Field	Type	Notes
name	string	Required. Unique in the workspace.
flavor	string	Required. e.g. gpu.a100, cpu.32.
auto_stop	string	30m, 2h or never. Default 1h.
image	string	Optional. Pinned on first start.

curl -s https://api.polnor.net/api/v1/compute \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name":"training-a100","flavor":"gpu.a100","auto_stop":"2h"}'

201 Created

{ "id": "cmp_8f21", "status": "stopped" }

Start a stopped instance. The image is persisted on first start for reproducible restarts.

Stop a running instance and release the underlying VM.

Delete an instance. Returns 409 if it is still running.

List warehouses and their state (stopped, starting, running, stopping).

Provision the warehouse VM with DuckDB and Spark sidecars. Takes roughly three minutes to reach running.

Run a query. Reads route to DuckDB, writes to Spark, automatically.

Request

curl -s https://api.polnor.net/api/v1/sql/execute \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"query":"SELECT count(*) FROM acme.sales.orders",
       "warehouse_id":"wh_4a2b"}'

200 OK

{
  "columns": ["count"],
  "rows": [[1284902]],
  "engine": "duckdb",
  "elapsed_ms": 412,
  "lineage": { "read": ["acme.sales.orders"] }
}

List databases (Iceberg namespaces) in the workspace.

List tables in a namespace, with row counts and current snapshot.

Return a table's column schema (name and Iceberg type).

{ "columns": [
  { "name": "id", "type": "long" },
  { "name": "amount", "type": "decimal(10,2)" }
] }

Create a job. Pass tasks to declare a DAG, or omit it for a single command.

{
  "name": "daily-pipeline",
  "tasks": [
    { "key": "ingest", "command": "python ingest.py" },
    { "key": "transform", "command": "python tx.py",
      "depends_on": ["ingest"], "max_retries": 2, "timeout_seconds": 1800 }
  ]
}

Update a job. tasks: null leaves the DAG intact; tasks: [] clears it. Editing never rewrites past runs.

Trigger a run. Subject to the max_running_runs quota.

{ "run_id": "run_4812", "status": "pending" }

Run status and timing. Statuses: pending, running, success, failed, cancelled.

Per-task rows for a run (empty for single-command jobs).

[ { "key": "ingest", "status": "success", "retry_count": 0 },
  { "key": "transform", "status": "running" } ]

Logs for a single task, streamed and archived.

Tables this run read and wrote, captured from each query. See Governance.

Cancel a run. Pending and running tasks are cancelled; the agent SIGTERMs then SIGKILLs live processes.

Log a metric. Stored as the latest value and as a step history.

{ "key": "auc", "value": 0.934, "step": 4200 }

Log many metrics in one call.

Full step / value / timestamp history for a metric.

Get a presigned URL to upload an artifact straight to your own S3.

Serving endpoints backed by a model version.

List serving endpoints with status, model version, QPS and latency.

Create an endpoint from a model version. Subject to the max_endpoints quota.

{ "name": "churn-prod", "model": "churn-v3", "version": 7 }

Daily roll-ups of the audit log.

GET /audit/analytics?from=2026-06-01&to=2026-06-12&group_by=day,action

List the current workspace's invoices (draft, sent, paid).

Usage and cost for a date range: provider cost, multiplier, and Polnor margin.

{ "data": [
  { "category": "compute_gpu", "provider_cost": 1842.50,
    "multiplier": 2.0, "total_cost": 1842.50 }
] }