Anveld Docs
Platform

Deployment

Packaging models and workloads for edge devices, fleets, and hardware targets.

Deployment is where Anveld tries to preserve the same project model you used during development.

What the Deployment Layer Handles

  • Deployment manifests
  • Artifact packaging
  • Model conversion helpers such as ONNX and TensorRT
  • Target selection for edge hardware
  • Runtime environment validation and security controls

Example Flow

anveld deploy --target jetson-fleet

That deploy step is expected to package the current project, attach the right manifests, and produce something the edge runtime can actually execute.

Edge Targets

The repository currently focuses on hardware and integration plugins including:

  • Jetson
  • ROS 2
  • Franka
  • Universal Robots
  • RealSense

The deployment pipeline stays generic while the hardware specifics live in plugins.

Training & Simulation

The local-first workflow for simulator-driven iteration, distributed training, and experiment tracking.

Cloud Platform

The managed service layer for fleets, training jobs, model registry, telemetry, and collaboration.

On this page

What the Deployment Layer HandlesExample FlowEdge TargetsRelated Reading