🛠️ TaskFlow Technical Support RAG Chatbot

This project is a Retrieval-Augmented Generation (RAG) based chatbot built using:

• 🧠 LangChain for chaining LLM + retrieval
• 🔍 HuggingFace Transformers for embeddings and LLM (FLAN-T5)
• 🗃️ Pinecone for vector database search
• 🌐 Streamlit for an interactive frontend

---

💻 Features

• Ask technical questions about TaskFlow (a fictional cloud-based project management tool)
• Uses a CSV of past support issues and responses
• Retrieves relevant solutions using vector similarity search
• Responds concisely using a language model

---

📂 Folder Structure

.
├── RAG_ChatBot.py           # Backend logic (RAG setup)
├── app.py                   # Streamlit frontend
├── data/
│   └── tech_support_vedant.csv  # Your dataset
├── .env                     # API keys (not committed)
├── .gitignore
├── sample_io.txt            # Example input/output
└── README.md

---

🛠 Setup Instructions

1. Install dependencies:

pip install -r requirements.txt

2. Set up your .env:

PINECONE_API_KEY=your_pinecone_key
HUGGINGFACEHUB_API_TOKEN=your_huggingface_token

3. Run the chatbot:

streamlit run app.py --server.fileWatcherType none

---

📘 Notes

• Make sure your .env and tech_support_vedant.csv are present.
• Do not commit .env — it contains sensitive keys.
• Dataset should include these columns:
  • Customer_Issue
  • Tech_Response
  • Issue_Category

---

👤 Author

Vedant Kasar

---

🔐 License

This project is for educational/demo purposes only. Keys should be kept secret and rotated if leaked.