Lyra was built as part of my senior Computer Science capstone, focused on scalable system design and intelligent matchmaking architecture.
Backend Architecture
FastAPI + PostgreSQL
Modular service-oriented structure with validated RESTful APIs for:
- Authentication (secure token-based access control)
- Profile management (user preferences, photos, metadata)
- Recommendation generation (compatibility scoring)
- Messaging workflows (real-time conversations)
Recommendation Engine
Compatibility-Driven Ranking
Combines explicit user preferences with behavioral signals. Algorithm approach:
- Constraint-based filtering (eliminate incompatible matches)
- Weighted scoring (preference alignment + interaction history)
- Ranking by compatibility percentile
Produces personalized recommendations balanced between diversity and relevance.
Frontend Implementation
React + TypeScript
Type-safe components with clean state management. Features:
- Dynamic feed updates with infinite scroll
- Optimistic interaction handling (swipe feedback before API response)
- Real-time messaging UI with typing indicators
- Smooth animations and transitions
Engineering Focus
This capstone demonstrates:
- Production-aware system design (scalability, maintainability)
- Intelligent algorithm implementation (recommendation systems)
- Full-stack capability (backend + frontend + database)
- Clean architecture principles (modular, testable code)