[lwb]

One possible value for $NEEDS_A_NAME is Infrastructure Engineer ... not to say that infra isn't always productized, because it sometimes is, but the instincts required just aren't the same when your customer is the team that works down the hall. Maybe Systems Engineer if you consider the type of engineer that optimizes for speed and other factors. Idk, names are hard.

[nitrogen]

SRE sort of fits that box, though I have to point out w.r.t. the OP's post that there is being scrappy (not building for edge cases that don't have confirmed ROI) and being reckless (ignoring any engineering concern whatsoever).

I've seen projects sink for having PMs too far on the reckless side. Frontend engineers who appear to be developing the day's hot feature end up building a massive platform on platform on (...ad nauseam) on Reactive React just to show a few forms, while the backend engineers get pushback every time they ask a question. The product sinks because it takes three weeks just to move a field from one form to a different form, you have to refactor the whole redux store, ...

These are the teams that go so far in the "manual" direction that they refuse to build support tools of any kind whatsoever, and then when the inevitable fires come nobody knows what's going on.

So product<->reliability might be often mutually exclusive just because that is what people are used to, but really engineers should be able to do both and know when to adjust, and so should the managers.