Cool topic! This project is really interesting. (From the title I thought it was a sort of generic cheminformatics article, but the fingerprints in question are x-ray absorption spectra, which are the targets of the model)

The really tough problem in x-ray absorption spectroscopy (and many other analytical techniques I guess) is model inversion, where you have some measured spectrum and you want to infer some unknown aspect of the structure and/or chemistry. Lots of interesting work to be done!

thanks for sharing your perspective on the topic, as a layperson myself. Hope people like you are successful in creating new techniques and moving the world forward to more scientific discoveries!