How are the types not totally screwed up in this kind of pattern? I get that this is JS, but since most of us at the level to be reading this post actually write TS, isn't it true that most of those transformations actually modify the type of the object, and therefor each would need to have a slightly different intermediary type?

When I actually have this problem, I usually define the input as an interface and the output as another interface(maybe extending the input interface).

Then I write a SINGLE function that calls those other functions, collects the data into variables, and then builds a new object of the new data type, and then returns that.

Maybe I'm missing the point here.