🛍️ Customer Support Q&A System

An AI-powered chatbot that ingests product reviews and metadata, stores embeddings in AstraDB/Pinecone, and answers user queries via a FastAPI backend and a simple HTML/JS frontend. It integrates Google embeddings, Groq LLM, and vector search for real-time recommendation and support.

---

🚀 Features

• Data Ingestion
  • Reads CSV of Flipkart product reviews
  • Transforms into LangChain Document objects

• Vector Store
  • Uses Pinecone or AstraDB for embedding storage
  • Configurable namespace/collection

• Embedding & LLM
  • Google text-embedding-004 embeddings (dim=768)
  • Groq or Google Gemini for answer generation

• Backend
  • FastAPI app (main.py) with /get endpoint
  • CORS-enabled, async endpoints

• Frontend
  • Simple HTML/JS (templates/chat.html)
  • Tailored CSS (static/styles.css)

• Config & Secrets
  • .env for API keys & endpoints
  • config/config.yml for thresholds & parameters

---

📂 Project Structure

.
├── .env                      # API keys & endpoints
├── main.py                   # FastAPI entrypoint
├── requirements.txt
├── setup.py
├── README.md
│
├── config/
│   └── config.yml            # Retriever & pipeline settings
│
├── data/
│   └── flipkart\_product\_review\.csv
│
├── data\_ingestion/
│   └── data\_ingest.py        # Ingestion pipeline class
│
├── exceptions/
│   └── exception.py          # Custom exception classes
│
├── logs/
│   └── logging.py            # Logger setup
│
├── models/
│   └── model.py              # Pydantic schemas
│
├── prompts/
│   └── prompt.py             # Prompt templates
│
├── retrievers/
│   └── retriever.py          # Vector store & retriever logic
│
├── static/
│   └── styles.css
│
├── templates/
│   └── chat.html
│
└── utils/
├── config\_loader.py      # Loads `config.yml`
└── model\_loader.py       # Loads embedding & LLM models

---

⚙️ Setup & Installation

1. Clone & enter project

   git clone https://github.com/yourname/customer_support_system.git
   cd customer_support_system

2. Create & activate virtual environment

   python -m venv venv
   # Windows
   venv\Scripts\activate
   # macOS/Linux
   source venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. Configure .env Create a .env file in the root with:

   GOOGLE_API_KEY=your_google_key
   GROQ_API_KEY=your_groq_key
   PINECONE_API_KEY=your_pinecone_key
   PINECONE_ENVIRONMENT=your_pinecone_env
   ASTRA_DB_API_ENDPOINT=your_astra_endpoint
   ASTRA_DB_APPLICATION_TOKEN=your_astra_token
   ASTRA_DB_KEYSPACE=your_keyspace

5. Edit config/config.yml Define your retriever and ingestion settings:

   retriever:
     top_k: 5
     score_threshold: 0.1
   
   ingestion:
     batch_size: 100
     collection_name: customer_reviews

---

🚦 Running the App

1. Start FastAPI

   uvicorn main:app --reload --port 8000

2. Open Frontend Navigate to http://localhost:8000 in your browser.

3. Interact • Ask product-related questions • Watch backend logs for ingestion & retrieval details

---

🔧 Customization

• Swap vector store: in retriever.py, switch between AstraDBVectorStore and Pinecone
• Change embedding model: in model_loader.py, pick another Google or OpenAI model
• Extend prompts: edit prompts/prompt.py for different Q&A personalities

---

🤝 Contributing

1. Fork this repo
2. Create a feature branch (git checkout -b feat/your-feature)
3. Commit & push (git commit -m "Add feature" + git push)
4. Open a PR for review

---

📄 License

MIT © [Your Name] Feel free to learn, modify, and share!