VICTOR - RAG-powered PDF Search System

A Retrieval-Augmented Generation (RAG) system for querying PDF documents using vector embeddings and LLMs.

Setup

1. Install Dependencies

conda create --name myenv python=3.11
conda activate myenv
pip install -r requirements.txt

2. Configure Environment Create a .env file in the root directory:

SITE_URL=http://localhost:3000
SITE_NAME=VICTOR
MILVUS_HOST=localhost
MILVUS_PORT=19530
MILVUS_COLLECTION=pdf_vectors
OPENROUTER_API_KEY=your_api_key_here
LLM_MODEL=alibaba/tongyi-deepresearch-30b-a3b:free
EMBEDDING_MODEL=BAAI/bge-m3
TOP_K=3
SEARCH_EF=64

3. Create Milvus collection

docker-compose up -d
cd scripts
python create_milvus_collection.py

4. install frontend

cd frontend
npm install

Run

Quick Start (Windows):

run_Victor.bat

Manual Start:

# Terminal 1 - Backend
cd api
uvicorn main:app --reload

# Terminal 2 - Frontend
cd frontend
npm install
npm run dev

Access the application at http://localhost:3000

Tech Stack

• Backend: FastAPI, ChromaDB, Sentence Transformers
• Frontend: Next.js 15, TypeScript, Tailwind CSS
• LLM: OpenRouter API (DeepSeek/Llama)