I provided a lot of the data currently in Apertium for Irish to Scottish Gaelic machine translation, which was taken from an ad hoc system for this language pair that I created in 2005. Since then, I've developed a more mature MT system that translates Scottish and Manx Gaelic into Irish, using statistics on the target side. I'd be happy to mentor a GSOC student (or anyone else) interested in porting that work over to Apertium.
More about me: for about 15 years I have been working on developing language technology for under-resourced languages around the world. I've developed corpora and spell checkers for many (20+) languages using a web crawler, statistical methods, and contributions from native speakers. I am primarily interested in the Celtic languages, and particularly Irish. I've created a grammar checker, monolingual and parallel corpora, and a semantic network for Irish, and do a lot of the localisation of open source software (Firefox, LibreOffice, KDE, ...)