The best way is to start doing *something*, even if you are not solving some particular problems and no one will want it right away.

Start getting comfortable with toolchain and programming, and very soon you'll see the pile of problems that are crying for a pair of good hands. So abandon your first experiment and start working on real problems this time.