If you've ever hit pandas' performance limits on large
datasets, you know the pain. Polars is blazing fast —
but rewriting all your code to learn a new API? No thanks.

So I built nitro-pandas.

What is it?

nitro-pandas wraps Polars under the hood while exposing
the exact same pandas interface.

Just change your import:

import nitro_pandas as npd # instead of pandas as pd

That's it. Same groupby, loc, iloc, read_csv —
powered by Polars.

Install

pip install nitro-pandas

Why it matters

• No learning curve if you know pandas
• Polars backend = way faster on big data
• Lazy evaluation for large files
• Auto-fallback to pandas for edge cases

GitHub

github.com/Wassim17Labdi/nitro-pandas

Would love feedback and contributions!

If you find it useful, a ⭐ on GitHub means
a lot and helps others discover the project!