project-akri/akri

Health v2 · maintenance only

Adoption not part of the score

1259 stars · 166 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Akri is a Kubernetes Resource Interface, a CNCF Sandbox project written in Rust, that exposes heterogeneous edge leaf devices (IP cameras, USB devices, GPUs, FPGAs) as schedulable resources in a Kubernetes cluster. It extends the Kubernetes device plugin framework to handle the dynamic appearance and disappearance of edge devices, discovering them via pluggable Discovery Handlers (ONVIF, OPC UA, udev) and scheduling workloads on nodes that have access to them.

Use cases

• expose USB cameras as Kubernetes resources
• schedule workloads on nodes with IP cameras using ONVIF
• discover OPC UA devices in an edge cluster
• make GPUs and FPGAs available to edge Kubernetes pods
• run Kubernetes on Raspberry Pi with attached sensors
• share leaf devices across pods in a cluster
• build custom discovery handlers for proprietary protocols

When to choose

• you run Kubernetes at the edge with heterogeneous devices too small to run K8s themselves
• you need devices dynamically discovered and workloads scheduled based on device availability
• you want a CNCF-backed, Rust-based solution with Helm deployment and Prometheus monitoring

When to avoid

• your devices are cloud datacenter hardware already served by standard device plugins
• you don't use Kubernetes
• you need a simple non-clustered device management tool

Facets

framework · maturity active

container-orchestration iot configuration-management scheduling iot cloud-computing infrastructure-as-code rust self-hosted edge-computing device-plugins leaf-devices onvif opc-ua udev helm cncf-sandbox usb-devices ip-cameras containers kubernetes linux docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="project-akri/akri")