Personal projects I've had:
A tool for cleaning up pirates MP3 files (very old).
A tool for executing arbitrary tasks via text message; this was particularly useful back in ~2005 when I had unlimited texting but no data plan.
A tool for visualizing my Nike+ running sensor data.
Simulation of a trading card game.
A recent project I haven't really started coding - a tool for organizing tasks in a dependency graph, which would be a superset of the feature of Google Inbox, and would likely integrate with email generally and maybe Trello or something as well.
Those aren't very data-oriented, but when I interview data scientists I still see that kind of thing as a big positive. If you can find / get a hold of data, that's a huge plus. If you have contacts at your university, that could be great, otherwise you can look for "open source" data. Ideally, you'd find something you're passionate about and then pursue the details from there, but it really depends on how luck you get as well.