🚀 Multilingual Agentic RAG System

A production-ready Retrieval-Augmented Generation (RAG) system with multilingual support and agentic architecture using open-source LLM models.

https://github.com/AdityaJ9801/Multilingual-Agentic-RAG/blob/5c18ef9cf0da327051cd6a106a7e4f29762abb14/Agentic_RAG_demo_video.webm

✨ Key Features

🌍 Multilingual Support

• ✅ 5 languages: English, Spanish, French, Chinese, Arabic
• ✅ Automatic language detection
• ✅ Multilingual embeddings
• ✅ Responses in query language

🤖 Agentic Architecture

• Router Agent: Routes queries to specialized handlers
• Retrieval Agent: Vector search and document retrieval
• Synthesis Agent: Generates responses using LLM
• Validation Agent: Fact-checking and quality validation
• Orchestrator pattern for agent collaboration

📦 Production Ready

• ✅ Fully tested (5/5 tests passed)
• ✅ Docker containerized
• ✅ Streamlit web interface
• ✅ REST API with FastAPI
• ✅ Vector database (Qdrant)
• ✅ Local LLM (Ollama)

📋 Prerequisites

• ✅ Docker & Docker Compose (v20.10+)
• ✅ Python 3.9+
• ✅ 8GB RAM minimum (16GB recommended)
• ✅ 20GB disk space
• ✅ Linux/macOS or Windows with WSL2

🚀 Quick Start (5 Minutes)

Step 1: Clone Project

git clone  https://github.com/AdityaJ9801/Multilingual-Agentic-RAG.git
cd Multilingual-Agentic-RAG

Step 2: Start Services

docker-compose up -d
sleep 60  # Wait for services to initialize

Step 3: Ingest Sample Data

bash scripts/ingest_sample_data.sh

Step 4: Install Streamlit

pip install -r streamlit_requirements.txt

Step 5: Launch Application

streamlit run streamlit_app.py

Step 6: Access Application

• 🎨 Streamlit UI: http://localhost:8501
• 📚 API Docs: http://localhost:8000/docs
• ✅ Health Check: http://localhost:8000/api/v1/health

📖 Usage

Via Streamlit Interface (Recommended)

1. Open http://localhost:8501
2. Go to Query tab
3. Enter your query in any language
4. Click Submit
5. View results with sources

Via API

Query the System:

curl -X POST "http://localhost:8000/api/v1/query" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What is machine learning?",
    "language": "en",
    "top_k": 5,
    "include_sources": true
  }'

Upload Documents:

curl -X POST "http://localhost:8000/api/v1/ingest" \
  -F "file=@document.txt"

List Documents:

curl http://localhost:8000/api/v1/documents

Check Health:

curl http://localhost:8000/api/v1/health

Configuration

Edit .env file to customize:

• OLLAMA_MODEL: LLM model to use (mistral, llama2, etc.)
• OLLAMA_TEMPERATURE: Response creativity (0.0-1.0)
• CHUNK_SIZE: Document chunk size in characters
• SUPPORTED_LANGUAGES: Comma-separated language codes
• EMBEDDING_MODEL: Multilingual embedding model

Architecture

┌─────────────────────────────────────────────────────────┐
│                    FastAPI Gateway                       │
│              (REST API, Request Validation)              │
└────────────────────┬────────────────────────────────────┘
                     │
        ┌────────────┴────────────┐
        │                         │
   ┌────▼─────┐          ┌───────▼──────┐
   │ Ingestion │          │ Query Engine │
   │ Pipeline  │          │ (Orchestrator)
   └────┬─────┘          └───────┬──────┘
        │                        │
        │    ┌────────────────────┼────────────────────┐
        │    │                    │                    │
   ┌────▼────▼──┐  ┌──────────┐ ┌▼──────────┐ ┌──────▼──┐
   │  Document  │  │  Router  │ │ Retrieval │ │Synthesis│
   │ Processor  │  │  Agent   │ │  Agent    │ │ Agent   │
   └────┬───────┘  └──────────┘ └───────────┘ └────┬────┘
        │                                           │
        │    ┌──────────────────────────────────────┤
        │    │                                      │
   ┌────▼────▼──────────┐              ┌───────────▼──┐
   │  Embeddings        │              │ Validation   │
   │  (Sentence-Trans)  │              │ Agent        │
   └────┬───────────────┘              └──────────────┘
        │
   ┌────▼──────────────┐
   │  Vector Database  │
   │  (Qdrant)         │
   └───────────────────┘

   ┌──────────────────┐
   │  LLM Service     │
   │  (Ollama)        │
   └──────────────────┘

Supported File Formats

• PDF: .pdf (via pdfplumber and PyPDF2)
• Text: .txt (UTF-8, Latin-1, CP1252)
• Markdown: .md
• JSON: .json
• CSV: .csv

API Endpoints

Method	Endpoint	Description
POST	/api/v1/ingest	Upload and process documents
POST	/api/v1/query	Submit queries and get responses
GET	/api/v1/documents	List ingested documents
DELETE	/api/v1/documents/{id}	Remove a document
GET	/api/v1/health	Health check
GET	/api/v1/agents/status	Agent status

🐛 Troubleshooting

Port Already in Use

docker-compose down
docker-compose up -d

Services Not Starting

docker-compose logs
docker-compose restart

Streamlit Connection Error

# Verify API is running
curl http://localhost:8000/api/v1/health

# Check Streamlit logs in terminal

No Documents Found

# Re-ingest sample data
bash scripts/ingest_sample_data.sh

Slow Responses

• Check Docker resources: docker stats
• Verify Ollama is running: docker-compose ps
• Reduce top_k parameter in queries

📁 Project Structure

multi_agentic_rag/
├── app/                    # Application code
├── scripts/                # Helper scripts
├── sample_data/            # Sample documents
├── streamlit_app.py        # Streamlit frontend
├── docker-compose.yml      # Docker configuration
├── requirements.txt        # Dependencies
├── INSTALLATION_GUIDE.md   # Installation steps
├── ARCHITECTURE.md         # System design
├── API_DOCS.md            # API documentation
└── README.md              # This file

🛑 Stopping Services

# Stop all services
docker-compose down

# Stop and remove volumes
docker-compose down -v