dev-arpit5462/augmented-research-assistant

sdk: streamlit

🔍 Augmented Research Assistant

A stunning dark-mode Streamlit RAG (Retrieval-Augmented Generation) application that allows users to upload documents and query information from them using cutting-edge AI technology.

✨ Features

• —Modern Dark UI: Professional dark theme with smooth animations and responsive design
• —Multi-format Support: Upload PDF, TXT, DOCX, and Markdown files
• —Intelligent RAG Pipeline: LangChain orchestration with LlamaIndex document processing
• —Google Gemini Integration: Powered by Google's latest Gemini API for inference and embeddings
• —Vector Search: ChromaDB for efficient document retrieval
• —Source Attribution: Display retrieved chunks to prevent hallucinations
• —Smart Caching: Embedding and query caching for improved performance
• —Chat Interface: Interactive chat with conversation history
• —Document Management: Upload, process, and manage your knowledge base

🏗️ Architecture

├── config/
│   ├── __init__.py
│   └── settings.py          # Configuration management
├── core/
│   ├── __init__.py
│   └── rag_chain.py         # LangChain RAG orchestration
├── services/
│   ├── __init__.py
│   ├── document_processor.py # LlamaIndex document loaders
│   └── vector_store.py      # ChromaDB + LangChain integration
├── utils/
│   ├── __init__.py
│   └── cache_manager.py     # Performance caching
├── app.py                   # Streamlit frontend
├── requirements.txt         # Dependencies
└── README.md               # This file

🚀 Tech Stack

• —Frontend: Streamlit with custom dark theme CSS
• —Orchestration: LangChain for RAG pipeline management
• —Document Processing: LlamaIndex for intelligent chunking and loading
• —LLM & Embeddings: Google Gemini API (free tier)
• —Vector Database: ChromaDB (local, no external dependencies)
• —Caching: Custom caching system for performance optimization

📋 Prerequisites

• —Python 3.8+
• —Google API Key (free from Google AI Studio)

🛠️ Installation

1. 1.Clone the repository:

git clone <repository-url>
cd augmented-research-assistant

1. 1.Create virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

1. 1.Install dependencies:

pip install -r requirements.txt

1. 1.Set up Google API Key:
2. 2.Visit Google AI Studio
3. 3.Create a new API key (free, no credit card required)
4. 4.Set environment variable:

   export GOOGLE_API_KEY="your_api_key_here"

Or create a .env file:

   GOOGLE_API_KEY=your_api_key_here

🎯 Usage

1. 1.Start the application:

streamlit run app.py

1. 1.Upload Documents:
2. 2.Use the sidebar to upload PDF, TXT, DOCX, or MD files
3. 3.Documents are automatically processed and indexed

1. 1.Ask Questions:
2. 2.Type questions in the chat interface
3. 3.Get AI-powered answers with source citations
4. 4.View retrieved document chunks for transparency

🔧 Configuration

Modify config/settings.py to customize:

• —Chunk Settings: CHUNK_SIZE, CHUNK_OVERLAP
• —Retrieval: TOP_K_RETRIEVAL
• —Model Parameters: TEMPERATURE, MAX_TOKENS
• —Caching: ENABLE_CACHING, CACHE_TTL

📊 Features Breakdown

Document Processing

• —LlamaIndex Loaders: Specialized loaders for each file type
• —Smart Chunking: Overlapping chunks for better context retention
• —Metadata Preservation: File names and chunk information maintained

Vector Storage

• —ChromaDB Integration: Local vector database via LangChain
• —Embedding Caching: Avoid recomputing embeddings for performance
• —Deduplication: Prevent duplicate document processing

RAG Pipeline

• —Context-Aware Retrieval: Top-k similarity search
• —Hallucination Prevention: Strict context-only responses
• —Source Attribution: Display retrieved chunks with relevance scores

User Experience

• —Dark Mode: Professional dark theme with gradients
• —Responsive Design: Works on desktop and mobile
• —Real-time Processing: Progress bars and status updates
• —Chat History: Persistent conversation memory

🚀 Deployment

Streamlit Community Cloud

1. 1.Push to GitHub:

git add .
git commit -m "Initial commit"
git push origin main

1. 1.Deploy on Streamlit Cloud:
2. 2.Visit share.streamlit.io
3. 3.Connect your GitHub repository
4. 4.Add GOOGLE_API_KEY in secrets management
5. 5.Deploy!

Local Production

streamlit run app.py --server.port 8501 --server.address 0.0.0.0

🎨 Customization

Styling

• —Modify CSS in app.py load_custom_css() function
• —Adjust colors, animations, and layout

Models

• —Change Gemini model in core/rag_chain.py
• —Adjust embedding model in services/vector_store.py

Processing

• —Modify chunk parameters in config/settings.py
• —Add new file type support in services/document_processor.py

🔍 Resume Highlights

This project demonstrates:

• —Modern AI Stack: LangChain + LlamaIndex + Gemini API integration
• —Production Architecture: Modular, scalable design with separation of concerns
• —Performance Optimization: Caching, efficient chunking, and vector search
• —User Experience: Professional UI/UX with dark theme and animations
• —Free Tier Usage: Entirely built on free APIs (no credit card required)

🤝 Contributing

1. 1.Fork the repository
2. 2.Create a feature branch
3. 3.Make your changes
4. 4.Add tests if applicable
5. 5.Submit a pull request

📝 License

MIT License - see LICENSE file for details.

🆘 Troubleshooting

Common Issues

1. 1.API Key Error:
2. 2.Ensure GOOGLE_API_KEY is set correctly
3. 3.Verify key is active in Google AI Studio

1. 1.Import Errors:
2. 2.Check Python version (3.8+)
3. 3.Reinstall requirements: pip install -r requirements.txt

1. 1.Performance Issues:
2. 2.Enable caching in settings
3. 3.Reduce chunk size for large documents
4. 4.Clear cache periodically

1. 1.Upload Failures:
2. 2.Check file size limits (10MB default)
3. 3.Verify file format is supported
4. 4.Ensure sufficient disk space

Support

For issues and questions:

• —Check the troubleshooting section
• —Review configuration settings
• —Ensure all dependencies are installed correctly

Built with ❤️ using LangChain, LlamaIndex, and Google Gemini API