For starters, what's your math background like? If you're interested in seriously pursuing ML theory, I'd recommend having at least 2 semesters of stats and 1 semester of linear algebra fresh under your belt. Don't short change yourself here, because any remotely advanced material will be incredibly frustrating (aka constant backpedaling) without the requisite math intuition.
One approach might be to tackle a MOOC like Andrew Ng's coursera offering (which isn't very math heavy) while simultaneously brushing up on your stats and linear algebra (mostly stats tbh). Even if you end up just focusing on implementation, I think this will be time well spent.
I'm a self trained dev (aka horrible person?) who's spent the last 8 years doing back end dev work - mostly C#, Java, Python and now Scala in everything from SV startups to a government research lab. Either I'm an outlier, or you're overestimating the importance of those cherished fundamentals (which can be picked up independently of a degree anyhow).
One approach might be to tackle a MOOC like Andrew Ng's coursera offering (which isn't very math heavy) while simultaneously brushing up on your stats and linear algebra (mostly stats tbh). Even if you end up just focusing on implementation, I think this will be time well spent.