According to Eugene Yan's latest post, knowing whether you need a research scientist, applied scientist, or ML engineer comes down to the work you actually need shipped. Research scientists focus on advancing the state of the art, producing new methods and models. Applied scientists translate those advances into products, while ML engineers are responsible for moving and maintaining models in real-world systems.
Yan notes the three roles require different skills and produce different deliverables, and he emphasizes that the boundaries can blur. He believes teams should choose based on the problem at hand rather than titles, and he offers his own distinctions as a personal framework.
Because this is a single personal take, the post does not contrast with competing definitions. But it gives practitioners and managers a clearer way to discuss what their teams are missing.