It’s a typical experience. Polars is fast, and Pandas is very slow and memory-hungry. It would be one thing if Pandas had a good API, but it doesn’t.