AI Powered Identity Risk Analyzer for Secure Access Decisions

We want a team to build a prototype that helps an organization decide how risky a given user, or a given access request, really is. Most companies grant access based on roles set up months or years ago and have very little visibility into whether a specific login or permission combination should be trusted in the moment. The tool would pull together signals like who the user is, what they normally do, what they're asking for now, when and where they're logging in from and whether their account is over permissioned compared to peers in similar roles, then produce a risk score along with a plain English explanation of why the score landed where it did. Students would build this as a web app with a backend service and a database. The analytics engine can start as a transparent rule based scorer and grow into a small ML model trained on synthetic or public IAM data. The explanation piece matters either way, since a security team won't trust a number that just appears with no reasoning behind it. The UI has to show which factors pushed the score up or down. Deliverables: a working prototype, source code, design and test documentation, a sample dataset and a final presentation. The bulk of the work is software design, backend and API development, data modeling, algorithmic logic and UI. The ML piece is one component, not the whole thing, which is why it sits on the CS side rather than DS, CY or IT.