"Make senior engineer sign off ai-assisted changes" sounds incredibly weird.

First thing that comes to mind is: reminds me of those movie where some dictatorship starts to crumble and the dictator start being tougher and tougher on generals, not realizing the whole endeavor is doomed, not just the current implementation.

Then again, as a former amazon (aws) engineer: this is just not going to work. Depending how you define "senior engineer" (L5? L6? L7?) this is less and less feasible.

L5 engineers are already supposed to work pretty much autonomously, maybe with L6 sign-off when changes are a bit large in scope.

L6 engineers already have their own load of work, and a fairly large amount of engineers "under" them (anywhere from 5 to 8). Properly reviewing changes from all them, and taking responsibility for that, is going to be very taxing on such people.

L7 engineers work across teams and they might have anywhere from 12 to 30 engineers (L4/5/6) "under" them (or more). They are already scarce in number and they already pretty much mostly do reviews (which is proving not sufficient, it seems). Mandating sign-off and mandating assumption of responsibility for breaking changes means these people basically only do reviews and will be stricter and stricter[1] with engineers under them.

L8 engineers, they barely do any engineering at all, from what I remember. They mostly review design documents, in my experience not always expressing sound opinions or having proper understanding of the issues being handled.

In all this, considering the low morale (layoffs), the reduced headcount (layoffs) and the rise in expectations (engineers trying harder to stay afloat[2] due to... layoffs)... It's a dire situation.

I'm going to tell you, this stinks A LOT like rotting day 2 mindset.

----

1. keep in mind you can't, in general, determine the absence of bugs

2. Also cranking out WAY MUCH MORE code due to having gen-ai tools at their fingertips...