Project AIBOM¶
An AI Bill of Materials (AIBOM) records what went into an AI workload run: the model, the dataset, the source code, the hardware, and how it performed. These projects collect that automatically on OpenShift and let you browse and compare the results.
| Project | What it does |
|---|---|
| Webhook | A Kubernetes admission webhook that instruments AI workloads and compiles an AIBOM custom resource when each run finishes. This is where AIBOMs come from. |
| CLI | oc aibom, a kubectl/oc plugin to list, filter, inspect, diff and compare AIBOMs from the terminal. |
| Console Plugin | An OpenShift Console plugin to browse, filter and compare AIBOMs in the web console, with telemetry charts. |
How they fit together¶
- Install the webhook and enable the namespaces whose workloads you want to track.
- Run Jobs as usual. Each one is instrumented without changes to its manifest, and an
AIBOMis created when it finishes. - Read the results with the CLI or the console plugin. Both use your existing Kubernetes RBAC.