Looking for advice on architecture of a music program

A few years ago I wrote a Haskell program for translating MusicXML scores (coming from music typesetting software like Sibelius, Dorico, or MuseScore) into MIDI events, in a way that provided custom interpretation of the scores and score objects. I did this to experiment with ways of adding flexible expression to a score that would otherwise sound stiff if played by the typesetting software itself or a DAW. There are some existing Haskell music packages but none of them allowed me to do enough ...