Nilometer — RAG-Based Academic Assistant for Computer Science Students

Independently designed and developed a full-stack web application using FastAPI and PostgreSQL, answering academic questions — FAQs, prerequisites, credit-hour calculations, and academic plans — for computer science students across 3 degree tracks. Combined semantic search (OpenAI embeddings + ChromaDB) with structured lookups for exact-accuracy calculations. Diagnosed and fixed a data-pipeline defect causing incorrect conditional credit-hour outputs, and added safeguards including per-IP rate limiting and conversation feedback logging. Containerized the application with Docker and deployed it to Azure Container Apps; currently live in a private pre-launch phase ahead of full release to Nile University CS students.