[andyjohnson0]

I kind of despair of keeping up to date with ML, at least to the extent that I might ever get current enough to be paid to work with it. I did Andrew Ng's Coursera specialisation a few years back - and I've worked through some of the developer-oriented courses, implemented some stuff. read more than a few books, read papers (the ones I might have a hope of understanding), and tried to get a former employer to take it seriously. But its seeming like unless you have a PhD or big-co experience then its very difficult to keep up to date by working in the field.

Notwithstanding the above, I'd agree with others here who suggest learning by doing/implementing, not reading papers.