Hot take:
If you need a massive data center to run a pattern matching software with the aggregated knowledge of several scientific fields, just to correctly write a series of config files, in order to exclude a couple of files from dependency optimization...
Maybe your infrastructure is just shit.
Maybe, just maybe, a large chunk of the web technology stack is plain garbage.
Have you considered this?