← All Projects
Complete

Sentinel AI

Vulnerabilities scanner for AI models. One command, five security checks, a single risk score.

View on GitHubVisit Site
PythonFastAPIReactPostgreSQLSecurityCLI
AI models are being deployed in hospitals, banks, and hiring systems - but there's no standard security check that happens before they go live. You wouldn't deploy software without scanning it for vulnerabilities. Why is nobody doing that for models?

That was the question that drove Sentinel AI at the hackathon. The analogy is VirusTotal - a tool you can point at a file and get a trust score back. We wanted the same thing for HuggingFace models: point it at any model ID, get a risk score back, know what's dangerous and why.