Self-hosted deployment

Runs inside your network

The full stack ships as a Docker Compose bundle. You bring the hardware; PraxisEdge brings the detector, OCR pipeline, bundled weights, dashboard, and cloud-compatible response schema.

What you get

The on-prem bundle is not a separate product surface. It is the same platform components arranged for customer-owned infrastructure.

Compose stack

The same Docker Compose deployment path used by the hosted plane, packaged for your network and hardware.

Data services

Postgres, Redis, ClickHouse, MinIO, NATS, Mosquitto, and the events services required by the platform.

Dashboard and APIs

The Next.js dashboard, REST API, streaming gateway, MQTT bridge, and the same response schema as cloud.

Pinned weights

Model artifacts ship with a manifest, and the API verifies the expected SHA-256 before loading weights.

Hardware minimums

Exact sizing depends on camera count, retention, and latency target. We do not publish a one-size-fits-all SLA here.

  • Docker and Docker Compose v2 are required.
  • NVIDIA GPU and NVIDIA Container Toolkit are recommended for production throughput.
  • CPU-only mode is supported for lower-volume deployments when latency requirements allow it.
  • Plan at least the repo-documented first-pull footprint, plus durable space for Postgres, ClickHouse, MinIO, and retained images.

On-prem specifics

Built for customers who need the LPR plane to stay under their operational control.

  • Private-network install runs entirely inside your network via the same Compose stack. Full air-gapped packaging is confirmed during scoping.
  • Weights are SHA verified at load from the bundle manifest.
  • Outbound access requirements are scoped per deployment so customer networking policies can be honored.

Scope an on-prem install

We'll size the deployment against your sites, cameras, and retention policy before quoting hardware or support terms.

Request a quote