I've plugged this several times here (just a fan) but chdb's DataStore is a "lazy" drop-in replacement for pandas dataframe. Pandas API with chdb performance. I don't use it myself (I do everything mostly in sql with either duckdb or chdb) but always found pandas more convinient than polars for quick and dirty data crunching (less typing and frankly more pythonic if you're used to slicing).