This feature set will include but is not limited to:

• Define the standard terms that will be used to through this AI Edge initiative
• Near Edge MVP - MLOps Pipelines
  • Fetch a trained model from a repository: Git, S3, Model Registry
  • Build a model into packaged into an immutable container with all of it's required dependencies
  • Test the model in a staging environment to verify that it is working as intended
  • Deploy the packaged model into a remote repository that will be made available at the Near Edge
• UX/UI - Provide a unified user experience for the user to minimize the requirement to understand all of the backend services required to build, deploy and monitor models running in near edge environments
• Near Edge MVP - Model Management in Near Edge Environments
  • Integration with ACM and ArgoCD to support deploying immutable model containers into near edge environments
  • Utilize ArgoCD to support the ACM Pull model for disconnected near edge environments
  • Utilize a GitOps to provide a source of truth for model deployment configuration
• Near Edge MVP - Observability
  • Monitor the operational metrics of my containerized model at the near edge.

• Near Edge MVP - CI/CD
  • Support for test a immutable monitor container
  • Updates to Pipelines required for the MLOps workflow

The first deliverable for the MVP will be a Pipeline and OpenShift AI Dashboard page incubating as feature in the odh-dashboard repository