This repository contains a complete Retrieval-Augmented Generation (RAG) workflow:

• Backend API in app/
• Frontend client in front/
• Notebook prototype in notebooks/

The system accepts only .txt and .pdf, does not persist uploaded files, and supports streamed answers.

• app/ FastAPI backend for RAG queries
• front/ React + Vite + TypeScript frontend
• notebooks/ Jupyter notebook prototype and local document workspace
• requirements.txt Python dependencies for backend/notebook

• Python 3.10+
• Node.js 24+ and Yarn
• A Hugging Face token

git clone <your-repo-url>
cd nlp_projects
pip install -r requirements.txt
cd front
yarn install
cd ..

uvicorn app.main:app --reload

Backend docs:

• Swagger UI: http://localhost:8000/docs

cd front
yarn dev

Open the URL printed by Vite (typically http://localhost:5173).

• Paste your Hugging Face token in the frontend (stored in browser localStorage)
• Upload a .txt or .pdf
• Enter a question (up to 1,000 characters)
• Toggle stream on/off and submit

Optional backend variables:

• CORS_ALLOWED_ORIGINS (comma-separated)
• EMBEDDING_MODEL
• LLM_MODEL
• CHUNK_SIZE
• CHUNK_OVERLAP
• NUM_RETRIEVED_DOCS
• HUGGINGFACE_API_KEY (fallback token if frontend token is not provided)

• Uploaded files are processed in-memory and are not stored by the API.
• Frontend build check:

cd front
yarn build

• Never commit .env or real API keys.
• Use .env.example as template and keep real values only in local env / deployment secret manager.
• Frontend sends token using Authorization: Bearer <token> at request time.
• If you suspect a token leak, rotate/revoke it immediately in Hugging Face settings.

If a secret was committed in git history, rewrite history before publishing:

git filter-repo --path .env --invert-paths
git push --force --all
git push --force --tags

• app/README.md
• front/README.md
• notebooks/README.md