Amazon certainly won't "fire you on the spot" in most cases; that gets them all sorts of trouble. However, if something got screwed up and your name is on it, I'm sorry to say it's just a matter of time before it comes up in a review cycle.
There are good folks at Amazon. Really. But as a whole, I don't see evidence of a company that even attempts to see the long-term value of people.
If you ever get slotted for a PIP (such a cute name...), it's probably better to look for new employment as soon as possible.
I mean that’s like literally the dictionary meaning of PIP :)
In my last firm we called it Period of Interview Preparation.
If no one can bring production down this is also a problem. If only one person can, better hope they don't get hit by a bus, or take a vacation, or rage quit.
Was that code merged without any review? If so, that is a recipe for disaster. Why no unit tests, integration tests, QA? What was the speed at which the outage was resolved? If a sane deployment strategy is in place, I'd hope that no big damage was actually done; if the error budget was burning fast, there should be some manual/automatic rollback.
tl;dr they may have been a key factor in the outage, but it seems like it is a symptom of a deeper hole in the process.