TraceScholar AI
An intelligent research assistant using Retrieval-Augmented Generation (RAG)
A full-stack AI research platform supporting web search, document uploads, automated summarization and contextual question-answering.
- Implemented document chunking, SQLite FTS5 retrieval, optional semantic embeddings and Reciprocal Rank Fusion (RRF) to generate source-grounded responses with citations.
- Built an interactive React frontend and FastAPI backend with SQLite-based session management, research history and configurable OpenAI/Groq integration.
- Developed a reproducible TF-IDF retrieval benchmark to assess document relevance, retrieval quality and ranking effectiveness using standard evaluation metrics.
- Python
- FastAPI
- React
- SQLite FTS5
- RAG
- OpenAI / Groq