ODH - Models as a Service with Policy Management

Our goal is to create a comprehensive platform for Models as a Service with real-time policy management.

📦 Technology Stack

• OpenShift: Kubernetes platform
• Gateway API: Traffic routing and management (OpenShift native implementation)
• Kuadrant/Authorino/Limitador: API gateway and policy engine
• KServe: Model serving platform
• React: Frontend framework
• Go: Backend frameworks

📋 Prerequisites

• OpenShift cluster (4.19.9+) with kubectl/oc access
• Kuadrant v1.4.2+ (ODH) or RHCL v1.3+ (RHOAI) - Required for MaaS v0.1.0+
• PostgreSQL database (for production ODH/RHOAI deployments)

⚠️ Important Version Requirements

Kuadrant 1.4.2+ Required (MaaS v0.1.0+)

MaaS v0.1.0 and later requires Kuadrant 1.4.2+ (ODH) or RHCL 1.3+ (RHOAI).

Why Kuadrant 1.4.2+ is required:

MaaS v0.1.0 requires the authorization header stripping capability added in Authorino v0.23.1 (shipped with Kuadrant 1.4.2) to protect user credentials from potential exfiltration to model backends.

Security Context:

When a user makes an inference request with their OpenShift token or API key, that credential must be validated by Authorino but should NOT be forwarded to model backends (whether internal KServe models or external providers). Kuadrant 1.4.2+ allows Authorino to:

1. Validate the incoming user credential (OpenShift token or MaaS API key)
2. Strip/replace the Authorization header before forwarding to model backends
3. Optionally inject model-specific credentials from Kubernetes Secrets (credentialRef) for ExternalModel resources

This prevents credential exfiltration where a malicious or compromised model service could capture and misuse user tokens.

Migration Notes:

• The deployment script (scripts/deploy.sh) automatically installs Kuadrant 1.4.2 for new deployments
• For existing deployments, upgrade Kuadrant/RHCL before upgrading to MaaS v0.1.0+

For detailed version compatibility, see Version Compatibility.

Database Required for Production

MaaS requires a PostgreSQL database for API key management. For production ODH/RHOAI deployments, you must create a Secret with the database connection URL before enabling modelsAsService.

See Database Prerequisites for details.

Note: The scripts/deploy.sh script creates a development PostgreSQL instance automatically.

🚀 Quick Start

Deploy Infrastructure

Use the unified deployment script for all deployment scenarios. The script installs prerequisites (policy engine, Gateway, PostgreSQL, Authorino TLS) and deploys maas-controller, which then deploys maas-api automatically via its Tenant reconciler:

# Deploy ODH (default)
./scripts/deploy.sh

# Deploy RHOAI
./scripts/deploy.sh --operator-type rhoai

# Deploy via Kustomize
./scripts/deploy.sh --deployment-mode kustomize

For detailed instructions, see the Installation Guide or the Deployment Options section below.

🛠️ Deployment Options

Basic Deployment

./scripts/deploy.sh [OPTIONS]

Key Options

Flag	Values	Default	Description
--deployment-mode	operator, kustomize	operator	Deployment method
--operator-type	odh, rhoai	odh	Which operator to install
--enable-tls-backend	flag	enabled	TLS for Authorino ↔ MaaS API
--disable-tls-backend	flag	false	Disable TLS backend
--namespace	string	auto	Target namespace
--verbose	flag	false	Enable debug logging
--dry-run	flag	false	Show plan without executing
--help	flag	-	Display full help

Advanced Options (PR Testing)

Flag	Description	Example
--operator-catalog	Custom operator catalog/index image	quay.io/opendatahub/catalog:pr-456
--operator-image	Custom operator image (patches CSV)	quay.io/opendatahub/operator:pr-456
--channel	Operator channel override	fast-3, stable-3.x

Environment Variables

Variable	Description	Example
MAAS_API_IMAGE	Custom MaaS API container image (passed to the Tenant reconciler via RELATED_IMAGE_ODH_MAAS_API_IMAGE)	quay.io/user/maas-api:pr-123
MAAS_CONTROLLER_IMAGE	Custom MaaS controller container image	quay.io/user/maas-controller:pr-123
OPERATOR_CATALOG	Custom operator catalog	quay.io/opendatahub/catalog:pr-456
OPERATOR_IMAGE	Custom operator image	quay.io/opendatahub/operator:pr-456
OPERATOR_TYPE	Operator type (rhoai/odh)	odh
LOG_LEVEL	Logging verbosity	DEBUG, INFO, WARN, ERROR

Note: TLS backend is enabled by default. Use --disable-tls-backend to disable.

Note: The policy engine is auto-determined based on operator type (rhcl for RHOAI, kuadrant for ODH/kustomize) and does not need to be set manually.

Deployment Examples

Standard Deployments

# Deploy RHOAI
./scripts/deploy.sh --operator-type rhoai

# Deploy ODH
./scripts/deploy.sh --operator-type odh

Testing PRs

# Test MaaS API PR #123
MAAS_API_IMAGE=quay.io/myuser/maas-api:pr-123 \
  ./scripts/deploy.sh --operator-type odh

# Test ODH operator PR #456 with custom manifests
./scripts/deploy.sh \
  --operator-type odh \
  --operator-catalog quay.io/opendatahub/opendatahub-operator-catalog:pr-456 \
  --operator-image quay.io/opendatahub/opendatahub-operator:pr-456

Minimal Deployments

# Deploy without TLS backend (HTTP for Authorino to maas-api)
./scripts/deploy.sh --disable-tls-backend

📚 Documentation

• Installation Guide - Complete deployment instructions
• MaaS API Documentation - Go API for key management
• MaaS Controller Documentation - Controller, Tenant reconciler, and subscription model
• Authorino Caching Configuration - Cache tuning for metadata and authorization

Online Documentation: https://opendatahub-io.github.io/models-as-a-service/

🤝 Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

📝 License

This project is licensed under the Apache 2.0 License.

📞 Support

For questions or issues:

• Open an issue on GitHub
• Check the Installation Guide for troubleshooting
• Review the samples for examples