This project implements a Document Search Assistant that allows users to upload documents, ask questions, and receive relevant answers using retrieval-augmented generation (RAG).

• Document Upload: Supports PDF, DOCX, and TXT formats.
• Text Extraction: Extracts and preprocesses text from uploaded documents.
• Document Search: Retrieves relevant parts of the document in response to user queries using a vector database.
• Answer Generation: Uses a generative model (Gemini API) to synthesize coherent answers based on retrieved content.
• User Interface: Provides a simple, interactive UI for file uploads and queries.

• Text Extraction: PyPDF2, pdfplumber, python-docx
• Embedding Generation: sentence-transformers (all-MiniLM-L6-v2)
• Vector Database: FAISS
• Generative Model: Gemini API
• Web Framework: FastAPI & React-Vite

• Clone the reporistory

• Make a .env file and put your gemini api key in following format: SECRET_KEY="Your Key"

• Backend:

      cd backend
      python3 -m venv venv
      source venv/bin/activate
      pip install -r requirements.txt
      uvicorn app.main:app --reload

• Frontend:

    cd frontend
    npm install
    npm run dev

• Add search history & document previews.
• Improve Retrieval Accuracy
• Allow users to upload and query multiple documents simultaneously.
• Implement cross-document search capabilities.