My complaints with FAANG have to do with perverse incentives that reward nonsensical decisions, poorly thought-out and over-engineered projects, grandiose documents, duplication of work, selective reporting of metrics, etc.
The few times I had a really good manager, a sane environment, and fulfilling work only lasted until the next reorg. It seems like most organizations are either stressful with a lot of adversarial behavior, or have almost nothing to do but depressing busywork. I also find the social aspect lackluster if not downright alienating. I feel at a dead end both in career growth and opportunities to learn on the technical side. I could roll the dice with another team change, but I'm not eager at the prospects.
Most of my work experience is in ML, but I don't want to box myself into that. I find the current hype around generative models insufferable, and the typical ML project today consists of somewhat sloppy Python and a lack of good engineering practices. I'm also tired of the increasingly long and opaque feedback loops (come up with an idea, wait for your giant model to retrain, hope that some metric goes up). I'm still passionate about some aspects (e.g., learning representations, knowledge grounding, sane ML workflows).
I hear that academia has similar issues (though again I mostly know about ML), and I imagine lots of industries have worse conditions than tech. I realize that sloppiness and politics are a fact of life, so I'm wary of falling into the "grass is greener" trap.