🤖 HR Chatbot with FAISS + GPT-4o

An interactive chatbot that answers HR policy questions using FAISS vector search and GPT-4o via GitHub-provided OpenAI token.

---

🛠 Features

• 💬 Chat interface using Streamlit
• 📁 HR documents embedded using sentence-transformers + FAISS
• 🔍 Context-aware question rephrasing
• 🤖 GPT-4o responses via GitHub Token (https://models.github.ai/inference)

---

📦 Installation

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/hr-chatbot.git
cd hr-chatbot

2. Install dependencies

pip install -r requirements.txt

3. Create a .env file

GITHUB_TOKEN=your_github_provided_token_here

---

📁 Folder Structure

├── faiss_index/              # Vector store (auto-generated)
├── hr-policies/              # Your HTML files (input)
├── hr-chatbot.py             # Main Streamlit chatbot app
├── hr-upload.py              # Script to build FAISS index from HTML
├── requirements.txt
└── README.md

---

🚀 Usage

1. Build FAISS Vector Index (run once)

Make sure your hr-policies/ folder contains all the HTML documents.

python hr-upload.py

2. Start the Chatbot UI

streamlit run hr-chatbot.py

You’ll be able to chat with the bot about HR policies in a Streamlit web UI.

---

🧠 Notes

• GPT-4o is accessed via GitHub’s models.github.ai inference endpoint.
• Your GITHUB_TOKEN must be valid and have access to that endpoint.
• Ensure faiss_index/ exists before running the chatbot.

---

📝 License

MIT – free to use, modify, and share.