Nonetheless I believe this is completely feasible.
Two possible approaches :
1. We'd need to build a program that would query relay(s) to get the events/notes published by one or multiple npubs and then generate the RSS XML files that then would be distributed by a webserver
2. A program that would be the webserver itself and that would request the relay when a URL is called to generate the XML content on the go
Second option would be the one I'd prefer