Introducing ChatDF
The Problem
You have a Parquet file with a million rows. You want to know the average value in column X grouped by column Y. You could fire up a Jupyter notebook, write some pandas code, wait for it to load, debug a typo, and eventually get your answer. Or you could just ask in English.
What ChatDF Does
ChatDF lets you chat with your data using natural language. Upload a Parquet file, ask questions in plain English, and get SQL-powered answers back instantly. The AI translates your question into SQL, runs it against your data, and returns the results — all in a conversational interface.
Key Features
- Natural language to SQL — ask questions in English, get precise SQL queries generated automatically
- Parquet file support — upload and query large Parquet files directly in the browser
- WebSocket streaming — responses stream in real-time as the AI generates them
- Rate limiting — built-in rate limiting to manage API costs
- Modern UI — clean React interface with syntax-highlighted SQL and formatted results
Tech Stack
The frontend is built with React and TypeScript, providing a responsive chat interface with real-time streaming via WebSockets.
The backend runs on FastAPI with Polars for lightning-fast data processing. Google's Gemini model handles the natural language to SQL translation, and DuckDB executes the generated queries against your uploaded data.
Try It
ChatDF is live at chatdf.daddaops.com. The source code is available on GitHub.