That said:
1) I ultimately felt powerless to make changes. Hiring people in to the service was (at the time) proving to be a massive uphill struggle. It just wasn't one of the "sexy" services to be working on, and the ops team didn't have a dedicated manager who could focus on that. It left about 90% of my work consisting of two things. Compliance work, and region builds. Even outside of that, a lot of the things that slowed us down were things like established tooling that had seen _zero_ proper business funding until later in my time there (I some are drastically improved), and this absolutely absurd obsession with re-inventing the wheel in every single team to do exactly the same things, even when the task required months of engineering time (I shudder at the thought of how much money got wasted paying engineers to re-do the same work, and re-discover the same bugs and problems.) I ultimately had no power to drive things forwards.
2) One manager left, and the replacement had absolutely no interest in the operational aspects of the platform. They seemingly only felt empowered to say "yes" to managers above them. Even if that meant their team was working 80 hour weeks. (the manager has since left and things have improved, so I'm told). One of the things that got to be absolutely insane was the amount of effort involved in launching a region. It needed dedicated developer time, and that particular manager just didn't give a damn. That manager's attitude and refusal to work on operational concerns were a big source of my frustration in the team. Looking around AWS, all I could see were other teams that were in a worse operational position, suffering serious levels of burn out, and I just decided I wanted nothing more to do with it.
Here's a big tip: If you make operations a priority, you can land features far faster. You don't keep tying up staff doing manual stuff or fighting fires.
I left AWS and joined Oracle to work on the Oracle Cloud Infrastructure, for a number of reasons. Not least of which was that a former manager from the service I worked for in AWS is running the Compute platform. I knew him, knew what he was like, and trusted him when he said one of the most important things to him was keeping operational burden of the platform down. I'd learned to trust his judgement in AWS, and it's something he's continued to deliver on here.
I'll be honest, this is one of the most enjoyable jobs I've ever had. Better pay, better benefits than AWS, great co-workers, senior directors with their priorities straight. Managers that politely but firmly insist you take time off if you ever have to do extra hours (and follow through with you if you don't.)