AI-Bot 🩺

An intelligent AI-powered medical chatbot designed to answer health-related queries using uploaded bio-medical encyclopedias and domain-specific knowledge. Built using Llama-3 (via Groq), Hugging Face Inference API, LangChain, Pinecone vector database, and Flask, this bot provides accurate, ultra-fast, and context-aware responses to user queries.

🔴 Live Demo: ✨️

---

🚀 Features

• Natural Language Medical Query Understanding: Capable of interpreting complex medical questions.
• Ultra-Fast Inference: Powered by Llama-3 via Groq for near-instant responses.
• RAG Architecture: Uses Retrieval-Augmented Generation to ground answers in verified medical texts.
• Cost-Efficient Embeddings: Utilizes Hugging Face Inference API for lightweight, cloud-based embeddings (No heavy local download).
• Vector Search: efficient document retrieval using Pinecone.
• Seamless Cloud Deployment: deployed live on Render.

🛠 Tech Stack

• Language: Python 3.10
• Framework: Flask
• Orchestration: LangChain
• LLM: Llama-3 (via Groq API)
• Embeddings: Sentence-Transformers (via Hugging Face Inference Client)
• Vector Database: Pinecone
• Deployment: Render

📘 Use Case

Uploaded "Gale Encyclopedia of Medicine" (bio-medical) books (as PDFs) so that users can interact with the bot to get accurate medical insights, references, and intelligent answers grounded in trusted data sources rather than generic AI hallucinations.

---

💻 How to run locally?

STEPS:

1. Clone the repository

git clone https://github.com/BleeGleeWee/AI-Bot.git
cd AI-Bot

2. Create a conda environment

conda create -n AiBot python=3.10 -y
conda activate AiBot

4. Install the requirements

pip install -r requirements.txt

6. Setup Environment Variables Create a .env file in the root directory and add your credentials. (Note: You need API keys from Groq, Hugging Face, and Pinecone)

PINECONE_API_KEY = "xxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
GROQ_API_KEY = "gsk_xxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
HUGGINGFACEHUB_API_TOKEN = "hf_xxxxxxxxxxxxxxxxxxxxxxxxxxxxx"

8. Ingest Data (Create Embeddings) Run this only once to process your PDFs and store vectors in Pinecone:

python store_index.py

10. Run the Application

python app.py

12. Access the Chatbot Open your browser and go to:

http://localhost:8080

---

🌐 Deployment (Render)

This project is currently deployed on Render as a Web Service.

Deployment Steps:

• Push to GitHub: Ensure your latest code (with requirements.txt and Procfile) is on GitHub.
• Create New Web Service: Log in to Render and connect your GitHub repository.
• Configure Settings:
  • Runtime: Python 3
  • Build Command: pip install -r requirements.txt
  • Start Command: gunicorn app:app
• Environment Variables: Add the following secrets in the "Environment" tab on Render:
  • PYTHON_VERSION: 3.10.12
  • PINECONE_API_KEY: (Your Key)
  • GROQ_API_KEY: (Your Key)
  • HUGGINGFACEHUB_API_TOKEN: (Your Key)
• Deploy: Click "Manual Deploy" -> "Clear build cache & deploy" to go live.

---

📂 Directory Structure

AI-Bot/
├── Data/                   # PDF files for knowledge base
├── src/
│   ├── helper.py           # Embedding & PDF loading logic
│   ├── prompt.py           # System prompts for Llama-3
├── templates/
│   └── chat.html           # Frontend UI
├── static/
│   └── style.css           # Styling
├── app.py                  # Main Flask application
├── store_index.py          # Script to ingest data into Pinecone
├── requirements.txt        # Project dependencies
├── Procfile                # Deployment command for Render
└── .env                    # API Secrets (Not committed to Git)

---

🤝 Contribution

We welcome contributions to improve the AI Medical Chatbot! Whether it's fixing bugs, improving documentation, or adding new features, your help is appreciated.

Steps to Contribute:

1. Fork the repository.
2. Clone your forked repo:

   git clone https://github.com/BleeGleeWee/AI-Bot.git

3. Create a new branch for your feature or fix:

   git checkout -b feature-name

4. Make your changes and commit them:

   git commit -m "Added a cool new feature"

5. Push to your fork:

   git push origin feature-name

6. Open a Pull Request (PR) on the main repository.

---

⭐ Support

If you found this project helpful or interesting, please consider giving it a Star! 🌟, helps others discover this project and motivates me to keep improving it.

---