You must not work with any kind of real world data if you think Excel is a horrible tool.
Pulling in 1000000000 rows of data you know absolutely nothing about to throw into your ML model and have it spit out some questionable result doesn't mean you actually work with real world data.
>a few hundred rows data science.
People that work with a few hundred rows are called "almost every office employee ever" and not data scientists. And you are naïve as hell if don't realize how powerful excel is for every random ass thing that happens in the real world.
I don't understand the use-case, Excel is horrible tool for anything else than stupid, a few hundred rows data science.