(some of it didn't apply to me personally, but I should highlight it anyway)
8) salary is competitive only to some point, it's bigger than most startups and mid-range companies but may be smaller than some large tech competitors can offer,
9) Amazon's motivation system (which unfortunately in place with the subsidiaries as well) is screwed in various ways:
9.1) reward loop may be anywhere from 1.5 to 2.5 years from the moment you did something cool to the moment you actually receive that extra stock (annual perf process with new stock grant given with half a year delay, and this stock grant only vests in a year),
9.2) there's a punishment if Amazon stock goes up, your pending unvested shares seem to be regularly re-evaluated at current market rate, and for a new grant you only get a delta of how much you deserve minus how much you're already getting (this may be zero for high performers if stock has grew by a lot),
9.3) there's also a punishment if Amazon stock goes down - you just get less that year, because any new compensation for the difference will vest only a year later;
based on my general observations, A9 probably does some tweaks to soften the effect of this stupid system, but it cannot completely dismiss the system;
10) probably due to the fact A9 is small and family-like, there's some notable variety in skills and personalities of managers in various groups; this, depending on your own personality and skills, may be beneficial for you or not so.
11) there are some gaps in common infrastructure such as source, build and deployment systems, which are outside of A9's control, and they either hurt productivity or require unwanted hacks or home-grown solutions, where instead the common infrastructure should get fixed.