Meridian State UniversityResearch Computing
Restricted Internal
AI System Card
Secure Research AI Cluster
| Document type | AI System Card | Version | 1.2 |
| Owner | Research Computing | Approver | AI Governance Committee + ISO |
| Data class | Confidential & Regulated | Visibility | Restricted Internal |
| Last reviewed | 2026-03-30 | Next review | 2026-06-30 |
1. System purpose & scope
An on-prem GPU environment for AI on sensitive and regulated research data. It runs open-weight models locally so that data never leaves the cluster boundary. Scope covers compute, storage, and the access pathway for approved studies.
2. Architecture
Isolated GPU nodes in a segmented network. Models run locally; there are no outbound calls to vendor inference APIs. Job scheduling and encrypted storage are managed by Research Computing.
3. Identity & access controls
Allocation
Per-project, time-boxed to the approved study period.
Authentication
Multi-factor authentication; least-privilege roles.
Segregation
Per-project data isolation; no cross-project access.
4. Security & privacy controls
- Network isolation and encrypted storage at rest
- Full audit logging and data-egress prevention
- Access recertification on a defined cycle
5. Governance boundary
Approved for Confidential and Regulated data under an IRB protocol and data-use agreement. General-purpose chat is out of scope — that belongs on Meridian GPT.
6. Review history
2026-03-30 · v1.2
Penetration test re-run; egress controls re-validated.
2025-09-12 · v1.0
Initial accreditation and governance approval.
Review & approval
System owner — Research Computing
Information Security Office
AI Governance Committee
Date approved