Perhaps also see my comment nearby about how "scientific computing" isn't really a narrow domain any more. It could add ammunition to your argument.