I may be a rare bird starting with R dataframes (still newbie+ level), then python polars (intermediate- ?). Frankly whenever I have to use pandas or df's in R I am not convinced that these are more intuitive/easier to master. I.e. I do not like the concept of row names.

Polars can be an overkill for small/medium dataset, but since I have been bitten by corrupted/badly formatted CSVs/TSVs I love the fact that Polars will throw the towel & complain about types/column number mismatches etc. And the fact that it can scale up to millions of rows on a modest workstation compensates the fact that sometimes one can spend hours finding a proper way to manipulate a dataset.