Asset Library in Hybrid Manager

The Asset Library in Hybrid Manager provides a central location for managing AI-related assets in your project. It replaces the prior Model Library and expands capabilities to include:

• AI model images used in Model Serving.
• Future support for additional AI artifacts (retrievers, embedding indexes, datasets).

You use the Asset Library to:

• Discover and manage NVIDIA NIM models.
• Integrate and manage private registries.
• Manage metadata and governance for model images.
• Support project-scoped model deployment and versioning.

How it fits in the Hybrid Manager architecture

• Asset Library overlays the Image Library provided by Hybrid Manager.
• All model images in Asset Library are backed by the project’s image repository and governance flow.
• Images can be deployed to KServe-based Model Serving endpoints within the same project.
• Asset Library is the central governance point for models used by:
• Gen AI Builder Assistants.
• AIDB Knowledge Bases (embedding models).
• Custom AI pipelines and APIs.

Asset Library ensures that all models used in your AI Factory workloads are controlled, auditable, and deployed securely within your project’s Kubernetes infrastructure.

Hybrid Manager considerations

• Repository configuration is scoped to each project. You can configure different private registries per project.
• The governance model follows Hybrid Manager controls, including image signing, provenance checks, and metadata tracking.

Typical use cases

• Deploy pre-integrated NVIDIA NIM models (LLMs, embeddings, OCR, rerankers).
• Add and manage organization-specific models from private registries.
• Control which model versions are available for different AI Factory workloads.
• Apply metadata and governance rules to track model usage and compliance.

Learn more

• How to deploy AI models in Hybrid Manager
• Manage repository metadata
• Full reference on the underlying concepts: Model Library Hub
• AI Factory Concepts