As a sidenote to your question, you may want to consider Data Engineering. It's not a sexy as ML, but it pays well and it's in high demand because somebody has

As a sidenote to your question, you may want to consider Data Engineering. It's not a sexy as ML, but it pays well and it's in high demand because somebody has to pipe all that data around so that the ML folks can do their thing. IMO it's much easier to go from a more traditional software development role into Data Engineering than into something as math-and-theory-heavy as ML because Data Engineering is based in how computers work and some knowledge of algorithmic scaling, not in heavy linear algebra/stats like ML.