When somebody searching for projects currently zanata doesn't filtering based on the translator known languages in first place. Instead of this - you have to select the project first, then the language - where do you get dozens of language that never even you gonna touch.
Based of the HU team feedback, it would be great to exchange this process with to set a language list within the user profile, and use that for tightened search overall in zanata. It would be much more simply to get the needed resource if the language would be the first then the project, or subproject. This also means that the remaining translation calculations on the selected resource shouldn't be based on all languages percentage, else only just the selected one, or two that the translator set up in his profile.
A simple solution (a first iteration if you will) could be defining a list of languages on your user profile, and only show the languages the translator actually care about. I don't know how many translators do translate more than one languages, but I am fairly sure that they are the exceptions, and not the norm. The current way of things is awfully similar to Transifex - which is annoying.