Think of an interesting non-software related domain that you enjoy and then use software to solve a problem in it. 10 years ago I scraped a run logger app backend for my running statistics and did data analysis on my GPS tracks that helped my training (things that Strava is doing today).

Don't make it into a product or publish it or do anything else that creates pressure on you. Just do it to solve your own problems. Be ok with the vast majority of those projects never being completed. The deal is to learn something and enjoy the pure act of creating something.