Agentic RAG System

Industry-grade Retrieval-Augmented Generation system with intelligent query routing, guardrails, and multi-source information retrieval.

Features

• Multi-document Upload: PDF, DOCX, TXT support
• Intelligent Routing: Automatically routes queries to:
  • RAG (uploaded documents)
  • LLM knowledge
  • Internet search (Serper API)
• Guardrails:
  • Input validation (malicious query detection)
  • Relevance scoring
  • Hallucination detection
• Source Attribution: Clear citations for all responses
• Streamlit UI: User-friendly interface
• Evaluation Framework: Comprehensive logging and metrics

Setup

1. Prerequisites

• Python 3.9+
• Groq API key (Get it here)
• Serper API key (Get it here)

2. Installation

# Navigate to project directory
cd C:\Users\ashfa\OneDrive\Desktop\My-Learning\Test

# Create virtual environment
python -m venv venv

# Activate virtual environment (Windows)
venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

3. Configuration

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key_here
SERPER_API_KEY=your_serper_api_key_here

You can copy .env.example and fill in your API keys:

copy .env.example .env

4. Run Application

streamlit run main.py

The application will open in your browser at http://localhost:8501.

Usage

1. Upload Documents: Click "Browse files" to upload PDF, DOCX, or TXT files
2. Process Documents: Click "Process Documents" to index them
3. Ask Questions: Type your question in the chat input
4. View Sources: Expand the "Sources" section to see where information came from

Architecture

Query → Input Validation → Query Analysis → Router
                                             ├─→ RAG Tool → Relevance Check → Response Synthesis
                                             ├─→ LLM Tool → Response
                                             └─→ Search Tool → Response
                                                              ↓
                                            Hallucination Check → Final Response

Configuration

Edit .env or config.py to customize:

• Chunk size and overlap
• Top-K retrieval
• Similarity thresholds
• Relevance/hallucination thresholds
• LLM model selection

Evaluation

View metrics by clicking "View Metrics" in the sidebar:

• Response times
• Source distribution
• Relevance scores
• Error rates

Logs are stored in ./logs/.

Project Structure

Test/
├── data/                    # Data persistence
│   ├── uploaded_docs/      # Original documents
│   └── chroma_db/          # Vector database
├── logs/                    # Application logs
├── src/                     # Source code
│   ├── agents/             # Agentic workflow (LangGraph)
│   ├── document_processing/ # Document loaders and chunking
│   ├── evaluation/         # Logging and metrics
│   ├── guardrails/         # Safety and quality checks
│   ├── llm/                # Groq client
│   ├── search/             # Serper client
│   ├── ui/                 # Streamlit components
│   └── vector_store/       # ChromaDB management
├── tests/                   # Unit tests
├── config.py               # Configuration management
├── main.py                 # Application entry point
└── requirements.txt        # Dependencies

Troubleshooting

Issue: "No module named 'src'"

• Solution: Ensure you're running from the project root directory

Issue: "API key not found"

• Solution: Check .env file exists and contains valid API keys

Issue: "ChromaDB error"

• Solution: Delete ./data/chroma_db/ and restart

License

MIT