I have no idea about pre-existent evolution simulation environments, but making your own is definitely feasible. That will simply require a solid programming background (perhaps physics as well, depending on what you have in mind). Based on your current background, I would suggest Python (or C/C++ if you really want maximum control over your hardware even though this will be much harder) and go with it.

There's this thing called google where you can insert text, type 'deep learning' and, after that, press the search button. You'll be amazed by how many results you'll get in a fraction of a second. Then you click on, say, a blog post, you start reading it. The first technical word that you see and don't know what it means you select it, right click it and press 'Search 'word' with google' and open whatever file that seems plausible. Do this for all unknown words on all posts, articles, papers, videos until your recursion ends. Also, don't forget to get your hands dirty!

Watch Karpathy's videos about it and you good to go!