Hi Robert, yes this has been the main concern for myself when implementing this same approach. I have integrated or seen it integrated in three different ways:

(1) like you said, we keep Python on server-side, the Excel user sends the Excel file and receives the outputs

(2) users have Python installed on their machines but do not use it beyond the scope of running the file (I’ve seen this used frequently in SQL/Excel heavy analytics teams)

(3) Excel users send the completed sheet to a given person/team, who runs it for them — essentially a manual, but infrastructure free version of (1)

I’ve seen 2 and 3 used a lot, 1 is less common (due to infrastructure in most cases)

Freelance ML engineer learning and writing about everything.

Freelance ML engineer learning and writing about everything.