Overview
Guides and reference material for the Anveld SDK, CLI, plugin system, deployment pipeline, and platform architecture.
Anveld is the developer platform for building, training, and deploying AI workloads for robots, drones, and edge devices.
The core idea is simple:
pip install anveld -> develop locally -> anveld deploy -> scale globallyWhat You Will Find Here
- Installation and first-run guidance for the local development loop
- The current SDK surface: protocols, data primitives, and infrastructure
- CLI command coverage for project setup, development, training, deployment, and operations
- Plugin architecture for trainers, environments, datasets, and hardware integrations
- Platform docs for simulation, deployment, and cloud services
- Deeper product and architecture material pulled from the repository source docs
Product Shape
Anveld treats the SDK as a small, stable core:
- 7 protocols define policies, trainers, environments, datasets, and multi-agent interaction.
- Infrastructure handles experiment tracking, deployment packaging, model conversion, and cluster orchestration.
- Algorithms, simulators, sensors, and hardware integrations are all plugins.
That separation is what allows the same project to move from local simulation to distributed training and then onto edge hardware without swapping frameworks in the middle.
Suggested Reading Order
- Start with Getting Started.
- Move to SDK Reference and CLI Reference.
- Use Plugin System when you are integrating your own trainer, simulator, or hardware target.
- Read Platform and Architecture when you need the deeper system model.