Web platform for clinical training, assessment and AI-assisted learning workflows, rebuilt into a production-ready React architecture.
Role
Focus
Stack
Contributed to a large clinical-education web platform combining structured learning, assessment and AI-assisted workflows. Worked across a React + TypeScript codebase to recover structure after vibe coding and turn an unstable AI-generated foundation into a maintainable production architecture.
The work was not a small feature, but a structural rescue after vibe coding: scattered logic, duplicate legacy paths, inconsistent multi-tenant scope, broken routes and weak separation between domains. The platform needed real cleanup before it could be treated as production software.
Delivered a production-oriented refactor with restored feature boundaries, safer access control, consistent tenant scoping, removed legacy dead ends, more predictable data flow and a codebase that could be extended without compounding earlier issues.
Reworked scattered AI-generated, vibe-coded code into clearer domain boundaries, reduced duplication and aligned the frontend with a predictable React project structure.
Audited and corrected role checks, tenant-scoped data flow and route protection so user data stayed isolated and predictable across the platform.
Removed dead routes, legacy hooks, stale screens and mock data that were blocking production readiness.
Hardened key flows, improved error boundaries and integrated runtime monitoring so production incidents could be detected and resolved quickly.
