"Pandas alternative" kind of undersells it -- it's drastically faster and supposedly has a much more intuitive interface. The limits of how long you can keep doing things entirely in-memory (and postpone the move to something like Spark) get higher and higher.

Corresponding blog post: https://pola.rs/posts/announcing-polars-1/

Pure love, thanks for sharing !