research-paper.pdf
chunked · embedded
Top-3 chunks
01 — AI-powered document Q&A
PDF Insight
Finding a specific answer inside a long PDF means scrolling and skimming for minutes.
A retrieval-augmented pipeline that extracts and chunks PDF text, embeds it, stores it in ChromaDB and answers questions grounded in the most relevant passages.
- PDF text extraction & semantic chunking
- all-MiniLM-L6-v2 sentence embeddings
- Persistent vector storage with ChromaDB
- Top-3 chunk retrieval via cosine similarity
- Context-grounded answers with Groq's LLaMA 3.3 70B
- Streamlit interface with custom CSS