I don't have solid advice on that, but if you are leaning towards a higher-risk approach you can do what I am doing: Work as a freelancer about 50% of your time

I don't have solid advice on that, but if you are leaning towards a higher-risk approach you can do what I am doing: Work as a freelancer about 50% of your time, do research in the spare 50%. The problem with that approach is that as a freelancer you usually need to constantly look for work, which I currently don't have to do, but that may change. The advantage of that approach is that your research is as free at it can possibly be: you are only accountable to yourself and your own goals. And you own your IP.