AI Governance Reference Architecture
Architecture
Reference architecture aligning NIST AI RMF outcomes to a platform control plane, implementing Govern/Map/Measure/Manage across delivery and runtime.
Project overview
This project documents a practical reference architecture that maps the four core RMF functions—Govern, Map, Measure, Manage—to platform services, approval workflows, evaluation infrastructure, and runtime controls. It is intended to help engineering teams translate policy into deployable controls.
Highlights
- Policy-as-code model approvals coupled to a model registry.
- Machine-readable AI bill of materials and data lineage.
- Evaluation harness, pre-deployment gates, and continuous monitoring.
- Per-model runtime controls, throttling, and automated triggers.