The research mentioned shows that the correlations are close to 0 even within similar work (there aren't many surgeons doing the equivalent of CRUD PHP webapps)

The research mentioned shows that the correlations are close to 0 even within similar work (there aren't many surgeons doing the equivalent of CRUD PHP webapps), which would imply that it's a filter that doesn't actually do you any good. I suppose you're arguing that the correlation is greater than 0 for an engineer and therefore it might be an effective filter. But how much greater? Particularly for industries where skill is important, using a marginally effective employment filter can significantly backfire.