VectorBrain

Overview

VectorBrain is a custom-built RAG (Retrieval-Augmented Generation) platform designed to run entirely on your own hardware. It allows users to upload sensitive documents and query them using natural language, powered by local LLMs for full data privacy and speed.

Core Features

• Fully self-hosted backend and frontend for local deployment
• WebSocket-powered real-time chat interface with token streaming
• Supports uploading and indexing `.txt` and `.csv` files
• Uses local GGUF LLMs via llama.cpp for inference (no external API calls)
• Token-based chunk retrieval with FAISS and vector search

Tech Stack

How to Run

• Create a virtual environment and install Python dependencies:

  python -m venv venv
  source venv/bin/activate
  pip install -r requirements.txt

• Start the backend server:

  uvicorn backend.main:app --host 0.0.0.0 --port 8000 --reload

• Start the frontend app:

  cd frontend
  npm install
  npm run dev

Source Code

View on GitHub