Knowledge Studio

A dark, premium Streamlit RAG app for lecture PDFs and document-grounded Q&A.

What it does

• Upload a PDF and build a persistent Chroma index
• Reuse saved indexes without rebuilding embeddings every run
• Ask grounded questions about the active document
• Generate a structured summary from retrieved chunks
• Switch between MMR and similarity retrieval
• Inspect retrieved source chunks and metadata
• Review simple analytics and feedback signals
• Search saved documents and export complete conversations

Stack

• Streamlit
• LangChain
• ChromaDB
• HuggingFace embeddings
• Groq
• PyMuPDF4LLM

Python version

Use Python 3.11 or 3.12.

Do not use Python 3.14 because Chroma and the Pydantic v1 compatibility layer can fail there.

Run locally

make setup
# create .env and add GROQ_API_KEY=your_real_groq_api_key
make run

Run make help to see all commands. The Makefile also provides install, compile, test, check, and clean targets. It uses python3.11 by default; override it with make setup PYTHON=python3.12.

Environment

Create a .env file in the project root and set:

GROQ_API_KEY=your_real_groq_api_key

Main pages

• Chat
• Documents
• Sources
• Analytics
• Settings

Notes

• PDF indexing is fully implemented.
• Other file types are surfaced in the UI for future extension.
• Destructive index deletion actions require confirmation.
• The UI styling is adapted from the uploaded knowledge_studio_ui.html design reference.