I do mobile release engineering. The unglamorous version: CI, the build pipeline, the app store submission dance, the part nobody thinks about until a launch is blocked. For about two years I've also been quietly doing a second, invisible job. Our pipeline is held together with workarounds, and most weeks something flaky breaks the release, and most weeks I notice it early on a Saturday or a 7 a.m. and fix it before standup so the team experiences a pipeline that "just works."
None of that is in my scope. None of it shows up in a review. It's the work that prevents embarrassment, which is reliably the work that nobody celebrates. I've raised the underlying fragility in planning four or five times. Each time it got acknowledged and deprioritized for feature work, because the pipeline kept working, because I kept making it work.
This week I stopped pre-empting. I did the release tasks that are actually mine, at the actual quality, on the actual timeline. I did not log in Saturday. I did not catch the flaky thing at 7 a.m. The flaky thing broke, the way it always does, and this time nobody invisibly absorbed it. The launch slipped two days.
Now there's a postmortem and the shape of it is "release was late." I'm in the frame as the owner of the pipeline that failed. The unspoken question in the room is what changed, and the honest answer is that I stopped donating my weekends to a problem the team voted four times not to fund.
Where I'm unsure: I knew this specific launch leaned on the pipeline. I knew the week I picked to stop was a week it would actually hurt. Part of me says that's exactly the point, the only time a hidden subsidy becomes visible is when you stop paying it during something that matters. The other part knows I could have stopped on a quiet week and made the same point with less blast radius, and I didn't, and I have to ask whether I let it slip on purpose.