Going against the grain here - you will have a much better time finding actual research job if that is your goal with a research degree. A Thesis based masters

Going against the grain here - you will have a much better time finding actual research job if that is your goal with a research degree. A Thesis based masters or PhD shows you can organize and present even incremental work in a useful way, moving the "state of the art" slightly vs. engineering around existing tools. It's 100% a subtle sliding scale - some engineering becomes research and much of research is engineering. ML Infra is just Infra, with a different set of needs and issues than say, Spring Boot infrastructure or something. Self taught here is fine, and honestly I'd trust a software engineer with basically no experience to handle ML infra better than an ML researcher.