Juan, thanks for sharing this article. I find your approach refreshing. I do see a use case which at some point I might implement in my Trainingsparkle startup: how you compare to yourself if you do a race multiple times. Interestingly Trainingsparkle also uses a lot techniques I read here (python, panda, haversine, ...) so it was a joy to read familiar code.

Just for info, the strava API streams do generate data which sometimes is a bit different from the gpx data (e.g. altitude)

https://www.instagram.com/trainingsparkle

Keep on running!