Hm I would suggest to just download the engine (and sdk) and play a little around for a few days first.
Maybee try to build a simple game eg a tetris clone, breakout ect to get a little the feel.
Then you can try to understand how the engine internally does the rendering, by following the source code trough the renedermanager ect.
Probably by then you already found something interesting to work on, be it some helper library, some improvements, a new postprocessing filter or something else
I concur with Empire. The only way you are going to have a good handle on what needs to be done and the proper way to do it would be to write some games with the engine and get some experience. If you’ve never done anything like this before then you have lots of stuff to learn and all of it is necessary for understanding the whole picture of why JME is the way it is and how it can be properly changed.
I’ve used quite a few rendering engines over the years and JME is one of the better architected ones… it’s why I picked it several years ago. That isn’t to say that there aren’t still some areas I want to change but these are not things to take lightly and certainly not without some holistic experience to back you up.