[iaw]

A few options :

1) Questions you want to answer... I had a question about the quality of two investment strategies so I'm in the process of solving them in R (where I'm proficient) and then later Python (where I'm learning)

2) Problems that have been posed... Euler problems may be the best for this http://projecteuler.net/problems

3) Finally, as others have said, real world problems that you or those you know have themselves. One thing I would like is an easy drop down wiki-style notes editor that has tree organization, that solution exists in multiple forms but not for free on OSX that I've found so it's something that I've considered building and open sourcing if I get back into compiled languages.

One anecdote : I share you're feeling about reading boring books, I'd rather do something real. I taught myself R by solving real problems one by one. An acquaintance is the opposite, he prefers to read the books, work through examples, play with them to be sure he understands them, and continue on. In the end he was able to show me a ton about optimizing code (as well as explain why certain things were slow). Just be aware that you might be skipping something intangible by skipping the book.