LangChain + OpenAI + Pinecone RAG Demo

A simple Retrieval Augmented Generation (RAG) application integrating LangChain, OpenAI, and Pinecone vector database, containerized with Docker.

Features

• LangChain: Orchestrates the RAG pipeline
• OpenAI: Provides embeddings (text-embedding-ada-002) and LLM (GPT-3.5-turbo)
• Pinecone: Vector database for semantic search
• Docker: Easy deployment and environment consistency

Project Structure

.
├── src/
│   └── main.py           # Main application code
├── data/                 # Data directory (for future use)
├── requirements.txt      # Python dependencies
├── Dockerfile           # Docker container configuration
├── docker-compose.yml   # Docker Compose setup
├── .env.example         # Environment variables template
└── README.md           # This file

Prerequisites

1. Docker & Docker Compose installed on your system
2. OpenAI API Key - Get it from OpenAI Platform
3. Pinecone Account - Sign up at Pinecone

Setup Instructions

1. Clone or Navigate to Project Directory

cd LangChain&RAG_demo

2. Configure Environment Variables

Create a .env file from the example:

cp .env.example .env

Edit .env and add your credentials:

OPENAI_API_KEY=sk-your-actual-openai-api-key
PINECONE_API_KEY=your-actual-pinecone-api-key
PINECONE_ENVIRONMENT=your-pinecone-environment
PINECONE_INDEX_NAME=langchain-demo

How to get these values:

• OpenAI API Key: Go to https://platform.openai.com/api-keys
• Pinecone API Key: Go to https://app.pinecone.io/ → API Keys
• Pinecone Environment: Found in your Pinecone console (e.g., us-east-1-aws)

3. Run with Docker Compose

docker-compose up --build

This will:

• Build the Docker image
• Install all dependencies
• Run the demo application

4. Run Without Docker (Optional)

If you prefer to run locally without Docker:

# Create virtual environment
python -m venv venv

# Activate virtual environment
# On Windows:
venv\Scripts\activate
# On Linux/Mac:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run the application
python src/main.py

What the Demo Does

The application demonstrates a complete RAG workflow:

1. Document Ingestion: Ingests sample documents about AI topics
2. Embedding Generation: Creates vector embeddings using OpenAI
3. Vector Storage: Stores embeddings in Pinecone
4. Semantic Search: Retrieves relevant documents for queries
5. Answer Generation: Uses GPT-3.5-turbo to generate answers based on retrieved context

Sample Output

Question: What is Artificial Intelligence?
Answer: Artificial Intelligence (AI) is the simulation of human intelligence
processes by machines, especially computer systems...

Architecture

User Query
    ↓
LangChain QA Chain
    ↓
OpenAI Embeddings (text-embedding-ada-002)
    ↓
Pinecone Vector Search (semantic similarity)
    ↓
Retrieved Context Documents
    ↓
OpenAI LLM (gpt-3.5-turbo)
    ↓
Generated Answer

Customization

Add Your Own Documents

Edit src/main.py and modify the sample_documents list:

sample_documents = [
    "Your first document text...",
    "Your second document text...",
    # Add more documents
]

Change LLM Model

Modify the ChatOpenAI initialization in src/main.py:

self.llm = ChatOpenAI(
    temperature=0.7,
    model="gpt-4",  # or "gpt-4-turbo-preview"
    openai_api_key=self.openai_api_key
)

Adjust Retrieval Parameters

Modify the retriever configuration:

retriever=self.vectorstore.as_retriever(
    search_kwargs={"k": 5}  # Retrieve top 5 chunks instead of 3
)

Troubleshooting

Issue: Pinecone Index Creation Fails

• Ensure your Pinecone API key is correct
• Check that you're using a valid environment/region
• Free tier Pinecone accounts have limitations on number of indexes

Issue: OpenAI API Errors

• Verify your API key is valid
• Check you have credits in your OpenAI account
• Ensure you're not hitting rate limits

Issue: Docker Build Fails

• Make sure Docker is running
• Try cleaning Docker cache: docker system prune -a
• Check your internet connection

Cost Considerations

• OpenAI: Charges for embeddings and LLM API calls
• Pinecone: Free tier available, paid plans for production
• Estimated cost for demo: ~$0.01-0.05 per run

Next Steps

• Load documents from files (PDF, TXT, etc.)
• Add a web interface with Streamlit or FastAPI
• Implement chat history and conversation memory
• Add document metadata filtering
• Deploy to cloud platforms (AWS, GCP, Azure)

Resources

• LangChain Documentation
• OpenAI API Reference
• Pinecone Documentation

License

MIT License - feel free to use this for learning and projects!