Before I knew very much music theory, I tried building a music generator. I learned TONS of music theory along the way, and the generator project sort of served as a testing ground for new concepts that I've learned, as well as a driving force that compelled me to learn about new aspects of theory in order to make improvements to the generator.

This is definitely not the most efficient way to learn music theory, but for me it always helps to have a project running alongside whatever I'm studying in order to immediately reap the benefits of new knowledge.