Probably not what you want, but I was looking for a while how I can get rid of Pandas, and for the most part ClickHouse (DBMS for analytics) can do almost every

Probably not what you want, but I was looking for a while how I can get rid of Pandas, and for the most part ClickHouse (DBMS for analytics) can do almost everything Pandas does (for my use case at least) and much more efficiently. It has very good support of array types, so it's pretty convenient for slightly nested data. Whatever you can't do completely in ClickHouse, you can export into Parquet/JSON/CSV/etc. and finish the analysis in Pandas or anywhere else. Just be aware that ClickHouse is still somewhat experimental and idiosyncratic DBMS, doesn't have some SQL goodies like window functions or B+Tree indexes, or pivots. They only added CTE support a week ago, I haven't even tried it yet.