Ask HN: Path from Software Development to Scientific Research?

There have been many stories of leaving academic STEM research to work in industry as software devs/data scientists/ML researchers, but I haven't heard many stories of people going the other way. What are viable paths for this? I’m not 100% tied to working in academia, but the goal would be to work on the heavily modeling / computational side of research for a scientific application and develop mastery in the application domain. This may come with pay cuts and years of studying, but I'm willing to make that tradeoff as long as it isn't too dire in the long term

Extra context, if helpful (sorry for the length):

I have a BSc in Math/CS from MIT + 5 years of software developer experience, including some ML work.

Since I’m not 100% sure what scientific problem domain I would like to work on / would reasonably be within reach, my gut feeling is that I should first do an interdisciplinary masters in something like "Computational Science" or "Scientific Computing" that would offer the following:

- Strengthen my foundations in mathematical modeling of different types of problems.

- Discover most fitting application domain. Do enough coursework in that application to start doing effective research.

- Experience working on research for said domain

My undergrad coursework covered foundational material in bio/chem/physics, but I’m willing to put in the work to fill the (possibly huge) gaps in the chosen domain. Over the past 1.5 years, I took a non-degree course in Quantum Computing, several MOOCs in physics, and read voraciously - unfortunately, I feel stuck trying to formalize this and find a path via a non-traditional route.