ME - 🤖 Generative AI-Powered Interactive App

An advanced interactive application leveraging cutting-edge Generative AI technologies to provide personalized responses and insights about Dhruv Kamalesh Kumar. The app employs an AI agent architecture with plans for further integrations, utilizes a Retrieval-Augmented Generation (RAG) function, and integrates with Pinecone for vector similarity search.

🌐 Live Demo: https://dhruvkamaleshkumar.streamlit.app/

📖 Table of Contents

• Features
• Architecture
• Technologies Used
• Installation
• Usage
• Configuration
• Project Structure
• Future Plans
• Contributing
• License
• Contact

✨ Features

• AI Agent Architecture: Utilizes an AI agent to handle complex tasks, reasoning, and decision-making for generating responses.
• Retrieval-Augmented Generation (RAG): Combines retrieved documents with generative models to produce contextually relevant answers.
• Pinecone Integration: Employs Pinecone as a vector database for efficient similarity search and retrieval of relevant information.
• Interactive Q&A: Users can ask questions and receive AI-generated responses based on Dhruv's professional background and experiences.
• Dynamic Projects Showcase: Fetches GitHub repositories dynamically and groups them based on AI-driven keyword analysis.
• Modern UI: Features a sleek, responsive user interface built with Streamlit, including navigation, profile highlights, and contact options.
• Resume Download: Allows users to download Dhruv's resume directly from the app.

🛠️ Architecture

The application's architecture is designed around an AI agent that orchestrates various components to provide intelligent and context-aware responses.

Components:

1. User Interface (Streamlit): Provides an interactive and user-friendly interface for users to engage with the app.
2. AI Agent (LangChain + OpenAI GPT-4):
   • Utilizes LangChain to manage interactions and reasoning with the language model.
   • Employs OpenAI's GPT-4 for generating coherent and contextually relevant responses.
3. Retrieval-Augmented Generation (RAG):
   • Retrieves relevant documents from a knowledge base using semantic search.
   • Enhances the generation process by providing the language model with pertinent context.
4. Pinecone Vector Database:
   • Stores embeddings of documents and enables fast similarity searches.
   • Integrates with the AI agent to supply relevant information for RAG.
5. GitHub Integration:
   • Fetches repositories using the GitHub API.
   • Analyzes repository data to group projects dynamically.

Data Flow:

• User Input: Users input questions or interact with the app.
• AI Agent Processing:
  • The agent interprets the input and determines the necessary actions.
  • If required, it retrieves relevant documents using Pinecone.
• RAG Functionality:
  • Combines retrieved information with the language model's capabilities to generate responses.
• Response Generation: The AI agent produces a final response, which is displayed to the user.

🚀 Technologies Used

• Streamlit: For building the interactive web application.
• OpenAI GPT-4: As the core language model for generating AI responses.
• LangChain: To structure the AI agent's reasoning and interaction with the language model.
• Pinecone: A vector database for efficient storage and retrieval of embeddings.
• GitHub API: For dynamic fetching and analysis of repositories.
• Retrieval-Augmented Generation (RAG): To enhance responses with relevant, retrieved information.

📥 Installation

Prerequisites

• Python 3.7 or higher
• OpenAI API key
• Pinecone API key and environment
• GitHub Personal Access Token (optional, for higher API rate limits)

Clone the Repository

git clone https://github.com/DB-25/ME.git
cd ME

Create a Virtual Environment

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

Install Dependencies

pip install -r requirements.txt

Set Up Environment Variables

Create a .env file in the root directory and add your API keys:

OPENAI_API_KEY=your_openai_api_key
PINECONE_API_KEY=your_pinecone_api_key
PINECONE_ENVIRONMENT=your_pinecone_environment
GITHUB_TOKEN=your_github_token  # Optional

Prepare Required Files

• Place your resume.pdf and memoji.png in the root directory.
• Ensure you have the ME.csv file with your documents for the RAG function.

🎮 Usage

Run the Streamlit app:

streamlit run app.py

Navigate to http://localhost:8501 in your web browser to access the app.

🌐 Or visit the live demo: https://dhruvkamaleshkumar.streamlit.app/

⚙️ Configuration

• OpenAI Model: The app uses gpt-4-1106-preview. You can change the model in the code if needed.
• GitHub Username: Update the github_username variable in the main() function to match your GitHub username.
• Grouping Keywords: Modify the group_keywords dictionary to customize how projects are grouped.
• AI Agent Settings: Adjust the temperature and other parameters in the ChatOpenAI initialization for different response behaviors.

📁 Project Structure

• app.py: The main application script containing the Streamlit app and AI agent logic.
• requirements.txt: Lists all Python dependencies.
• ME.csv: Contains documents used in the RAG process.
• resume.pdf: Dhruv's resume available for download.
• memoji.png: Profile image displayed in the app.

🔮 Future Plans

• Integrate Additional AI Models: Plan to incorporate other models like GPT-4 with plugins for enhanced capabilities.
• Expand RAG Functionality: Improve document retrieval methods and integrate more extensive knowledge bases.
• Enhance AI Agent: Implement more complex reasoning and decision-making processes within the AI agent architecture.
• Add Chat History: Allow users to view and interact with past conversations.
• Deploy to Cloud Platforms: Host the app on platforms like AWS or Heroku for broader accessibility.
• User Authentication: Implement login functionality for personalized experiences.

🤝 Contributing

Contributions are welcome! Please open an issue or submit a pull request for any enhancements, bug fixes, or suggestions.

📄 License

This project is licensed under the MIT License.

📬 Contact

• Email: dhruvbaradiya@gmail.com
• LinkedIn: linkedin.com/in/dhruvkamaleshkumar
• GitHub: github.com/DB-25