My PhD was in image processing - never really got to use it in any day job, but I published an iPhone app that back in the day that would read a sudoku puzzle and solve it. That led to a few side gigs and I’ve done off and on consulting jobs in the image processing space over the years.

Most problems tend to be amenable to “classic” techniques (eg OpenCV) and if you can get involved early enough there are a lot of affordances that can be added to make things a lot simpler (e.g registration marks).

I’ve also got a small YouTube channel that brings in a bit of revenue from ads and sponsorship.