README.md

---
title: Amazon Analytics Chatbot
emoji: 📦
colorFrom: blue
colorTo: green
sdk: docker
app_port: 8501
pinned: false
license: mit
short_description: RAG + SQL chatbot for Amazon seller analytics (demo data)
---

# 📦 Amazon Analytics Chatbot

A Streamlit chatbot (Docker-backed) that answers questions about Amazon seller
analytics using **SQL templates** for quantitative queries and **FAISS semantic
search (RAG)** for qualitative ones.

## 🏗️ Architecture

- **SQL Engine** — SQLite + SQLAlchemy, template-based query generation
- **RAG** — `sentence-transformers/all-MiniLM-L6-v2` embeddings + FAISS index
- **LLM** — Hugging Face Inference API (default: `Qwen/Qwen2.5-7B-Instruct`)
- **UI** — Streamlit with a custom dark theme, served via Docker

## 📁 Files

```
Dockerfile            ← Build & run instructions
requirements.txt      ← Python deps
src/
├── streamlit_app.py  ← Streamlit UI (entry point)
├── rag_core.py       ← RAG + SQL engine
├── company_data.db   ← SQLite database (demo data)
├── rag.index         ← FAISS vector index
└── rag_chunks.parquet← Chunk metadata
```

## 🔑 Secrets

In **Settings → Variables and secrets → New secret**, add:

- `HF_TOKEN` — your Hugging Face access token (read scope)

Optional:

- `HF_MODEL` — override default model, e.g.
  `meta-llama/Llama-3.2-3B-Instruct`

## 💡 Example Questions

- "Total revenue in 2023 Q1"
- "Monthly sessions trend last 30 days"
- "Top search terms by spend"
- "2024 H1 B2B revenue by state"

> The included database contains **demo / synthetic data** only.

## 📝 License

MIT