ragFlow

An asynchronous RAG (Retrieval-Augmented Generation) pipeline

---

Prerequisites

• Python 3.10+
• Docker Desktop installed and running
• A Groq API key from console.groq.com

---

Getting Started

1. Clone the repository

git clone https://github.com/scewdoosh/ragFlow
cd ragFlow

2. Start the Docker containers

Run the following command in the root folder to pull and start Qdrant and Valkey:

docker compose up -d

3. Create a virtual environment

Windows

python -m venv venv
venv\Scripts\activate

Mac / Linux

python3 -m venv venv
source venv/bin/activate

4. Install dependencies

pip install -r requirements.txt

5. Set up environment variables

Create a .env file in the root folder:

GROQ_API_KEY=your_groq_api_key_here

6. Run the server

Navigate one level above the project folder and run:

cd ..
python -m ragFlow.main

7. Start the worker

Open a new terminal, activate the venv, navigate one level up, then run:

Windows

venv\Scripts\activate
cd ..
rq worker --worker-class rq.SimpleWorker

Mac / Linux

source venv/bin/activate
cd ..
rq worker --worker-class rq.SimpleWorker

Note: The more workers you run, the more requests can be processed concurrently.

---

Usage

Open http://localhost:8000/docs in your browser to access the interactive API docs.

API Routes

Method	Endpoint	Description
GET	/hello	Health check confirms the server is running
POST	/upload_pdf	Upload a PDF by providing its local file path
POST	/chat	Submit a query returns a job_id
GET	/response	Fetch the answer using the job_id

Example Workflow

1. Upload your PDF

   • Hit POST /upload_pdf with your local PDF path, e.g. C:/folder/filename.pdf (Windows) or /home/user/filename.pdf (Mac/Linux)
   • Wait for the success message before querying

2. Ask a question

   • Hit POST /chat with your query
   • Copy the job_id from the response

3. Get your answer

   • Hit GET /response with the job_id
   • The answer will include the relevant page numbers from the PDF

---

Tech Stack

• FastAPI REST API
• RQ (Redis Queue) Async job processing
• Valkey Redis compatible message broker
• Qdrant Vector database
• HuggingFace Embeddings sentence-transformers/all-MiniLM-L6-v2
• Groq LLM inference