This is a good idea.
I have no idea what people would want or what I could make that people would want or what open problems are there that I could make. Every idea I have someone else already has a better solution.
I keep trying and put it in github but it's mostly for me; no one looks at it.
The Julia people are complaining that their language is great but it is not being picked up because their standard libraries are missing a ton of functionality.
Start knocking out some standard library functionality.
Sure, you may not be interested in Julia. But there are plenty of other projects out there that are understaffed. IPython Notebook needs developers. Octave needs somebody to write a good front end. It goes on and on.
The advantage is that you don't have to be super creative (compare Julia's libraries to Python, and start implementing something that isn't yet in Julia), nor predict the future (meaning, you can make a brand new X, but if the world goes to Y, you will never get picked up). What you write will be used by (at least) thousands, and you will have to write production quality code to get your pull requests accepted. I'd be impressed with anyone that did that, even if I had no interest/need in the project that they contributed to.
Whatever you made will probably be something pretty sweet because you chose something close to your heart, not something that was simply 'unsolved' or 'needed doing'.
Employers will love that you chose to challenge yourself to build new skills as a hobby, and because you chose something that you are passionate about, you'll probably also have a good amount of enthusiasm while you talk about it, which is equally important.