ContextGPT

ContextGPT is an AI-powered document retrieval and answering system. It enables users to upload documents, retrieve relevant context, and generate AI-powered responses using Ollama. The system uses ChromaDB as a persistent vector database for efficient document search and retrieval.

Features

• 📂 Upload PDFs and store document embeddings for retrieval.
• 🔍 Retrieve relevant context from stored documents using ChromaDB.
• 🤖 Generate AI-powered answers using an Ollama-backed inference service.
• 🖥️ Simple HTML frontend for seamless interaction.
• 🐳 Fully containerized with Docker for easy deployment.

---

Deployment Instructions

1️⃣ Clone the Repository

git clone https://github.com/nelabdiel/contextgpt.git
cd contextgpt

2️⃣ Install Dependencies

Ensure Python 3.11+ is installed, then run:

pip install -r requirements.txt

3️⃣ Start Services Without Docker

You can manually start each service by running the following in separate terminals:

cd document_service && python document_service.py

cd retrieval_service && python retrieval_service.py

cd inference_service && python inference_service.py

cd frontend_service && python frontend_service.py

Now, open your browser and go to http://localhost:5000

4️⃣ Start With Docker (Recommended)

Ensure Docker and Docker Compose are installed, then run:

docker-compose up --build

This will build and start all services in the background.

5️⃣ Access the Frontend

Once running, open your browser and go to: 👉 http://localhost:5000

---

🛠️ Project Structure

/project-root
│── docker-compose.yml        # Manages all services in Docker
│── frontend_service/         # Handles UI & routes requests to backend services
│   ├── Dockerfile
│   ├── frontend_service.py
│   ├── templates/
│   │   ├── index.html        # Updated HTML frontend
│── document_service/         # Handles PDF uploads & vector storage
│   ├── Dockerfile
│   ├── document_service.py
│   ├── uploads/              # Stores uploaded PDFs
│   ├── chroma_db/            # Persistent ChromaDB storage
│── retrieval_service/        # Retrieves relevant text from vector DB
│   ├── Dockerfile
│   ├── retrieval_service.py
│── inference_service/        # Calls Ollama for AI-generated answers
│   ├── Dockerfile
│   ├── inference_service.py
│── requirements.txt          # Shared dependencies
│── README.md                 # Project documentation

---

🖥️ API Endpoints

📂 Document Service (Port 5001)

• Upload PDFs → POST http://localhost:5001/upload
• Check status → GET http://localhost:5001/status

🔍 Retrieval Service (Port 5002)

• Retrieve relevant text → POST http://localhost:5002/retrieve

🤖 Inference Service (Port 5003)

• Generate AI response → POST http://localhost:5003/ask

🖥️ Frontend Service (Port 5000)

• Main UI → GET http://localhost:5000/

---

🔄 Stopping and Restarting

To stop all containers:

docker-compose down

To restart with fresh logs:

docker-compose up --build

For manual (non-Docker) deployment, stop services using CTRL+C in each terminal window.