A Helm chart for deploying Tekton pipelines for AI/ML model operations and GitOps workflows.

This Helm chart provides a comprehensive set of Tekton pipelines and tasks for:

• Model validation and testing
• Image scanning and signing
• SBOM generation
• Model deployment to Kubeflow Pipelines
• S3-based model retraining workflows

• GitHub Integration: Automated triggers for GitHub events
• Model Operations: Support for model scanning, validation, and deployment
• Security: Image scanning, signing with cosign, and SBOM generation
• Kubeflow Integration: Deploy models to Kubeflow Pipelines (KFP)
• S3 Integration: Model lookup and retraining from S3 storage

Deploy the pipeline to your OpenShift/Kubernetes cluster:

helm template . | oc apply -f- -n nine-thousand-models

Key configuration values in values.yaml:

• USER_NAME: Default user for operations
• git_server: Git server hostname
• cluster_domain: Cluster domain
• model_git_url: URL to the models repository
• image_scan: Enable container image scanning
• image_signing: Enable image signing with cosign
• generate_sboms: Generate Software Bill of Materials

• Security: Image scanning, signing, SBOM generation
• Model Operations: Model scanning, metadata updates
• Deployment: Build model containers, deploy to KFP
• Git Operations: Fetch commit hash, apply features

• github-pipeline: Main GitHub-triggered pipeline
• model-pipeline: Model-specific operations
• kfp-deploy-pipeline: Kubeflow Pipelines deployment
• s3-model-retrain-pipeline: S3-based model retraining

• GitHub webhooks for automated pipeline execution
• Event listeners for different pipeline types
• S3 event triggers for model retraining

• OpenShift/Kubernetes cluster
• Tekton Pipelines operator
• Appropriate RBAC permissions
• Cosign for image signing

Current version: 0.0.6

For more information, visit: https://rhoai-mlops.github.io/mlops/