DOCXBUD

Architected an agentic RAG pipeline using LangGraph with conditional routing across four study modes (Q&A, Flashcards, Quiz, Summary), implementing LLM-based multi-query expansion with deduplicated FAISS retrieval to improve cross-topic answer accuracy.

Engineered a semantic search pipeline using sentence-transformers embeddings and a FAISS vector store with recursive text chunking, grounding all LLM responses strictly in uploaded lecture material via LangChain retrieval

Developed a production-ready REST API using FastAPI with JWT authentication, bcrypt encryption, rate limiting, and MongoDB Atlas persistence for user sessions, document metadata, and resumable chat history

Upload a document and ask questions about your document(even multi query questions) and retrieve quick answers, your chats will be saved as you will have to login. Useful features for your document like quiz, flashcards, summary. Includes a Rate limiter as well.