localRAGvault

A fully local, privacy-first Retrieval-Augmented Generation (RAG) pipeline. This tool allows you to ingest documents and query them locally without relying on external APIs.

Tech Stack

• Generation Model: gemma4 (via Ollama)
• Embedding Model: embeddinggemma / nomic-embed-text (via Ollama)
• Vector Database: PostgreSQL 16 + pgvector
• Backend: Python (managed by uv) + FastAPI
• Frontend: Streamlit (Prototype) -> React/Vite (Stable)

Prerequisites

Before running this project, ensure your local environment meets the following requirements:

1. Ollama (v0.21.0+)
   • Models required: ollama pull gemma4 & ollama pull nomic-embed-text
2. PostgreSQL (v16+)
   • Must have the pgvector extension installed.
3. uv (v0.11.6+)
   • Used for Python dependency management.

Installation & Setup

1. Clone the repository and navigate to the directory:

   git clone https://github.com/kemalbsoylu/localRAGvault.git
   cd localRAGvault

2. Sync the Python environment using uv:

   uv sync

3. Configure your environment variables: Create a .env file in the root directory and add your Postgres credentials:

   DB_NAME=localragvault
   DB_USER=kemal
   DB_PASSWORD=your_secure_password
   DB_HOST=localhost
   DB_PORT=5432

Running the Application

This project currently uses a split architecture for development. You will need to run the backend and the frontend in separate terminal windows.

1. Start the FastAPI Backend

uv run uvicorn core.api:app --reload

The API will be available at http://127.0.0.1:8000

2. Start the Streamlit Frontend Open a new terminal window and run:

uv run streamlit run ui/app.py

The UI will automatically open in your browser at http://localhost:8501

Project Structure

• core/: Contains the FastAPI application, database connections, and RAG utility functions.
• ui/: Contains the Streamlit frontend prototype.
• uploads/: Local storage for ingested documents.
• tests/: Test scripts for verifying API endpoints.