Click to Zoom
Lyra was built as part of my senior Computer Science capstone, sponsored and developed under my leadership as the primary engineer.
Architecture & Deployment
- Full-Stack Implementation: Designed and built the entire system using FastAPI for a fast, modern backend and React (with TypeScript) for a highly responsive, type-safe user interface.
- Production Infrastructure: Containerized and deployed services using Vercel for the frontend and Render for the API server, backed by a persistent PostgreSQL database.
- Sole Contact: Served as the sole technical point of contact for the project sponsor, managing requirements gathering, timeline commitments, and milestone delivery.
Weighted Matchmaking Engine
- Algorithmic Scoring: Designed and built a custom weighted matchmaking engine in Python. The algorithm computes compatibility scores across multiple user preference dimensions.
- Intelligent Recommendations: Integrated OpenAI API to power a natural recommendation engine that identifies matches based on conversational profiles and behavioral cues.
- Guardian AI: Incorporated safety protocols and automated moderation (Guardian AI) using LLMs to scan and maintain community guidelines within messaging interfaces.
Database Design
- Optimized PostgreSQL schemas to handle complex user preferences, swiping histories, and real-time conversation records.
- Implemented robust index structures and transactional isolation levels to ensure data integrity during simultaneous reads and writes.