Site Reliability Engineer

Know your product is down before your customers tell you, and know exactly what it cost.

Your Site Reliability Engineer measures the one thing that matters: whether people could actually do what they came to do. Not server graphs, not a green dashboard. It sets a target you choose, watches how fast failures are eating into it, and only interrupts a person when the rate genuinely threatens the month.

It runs the incident, so the incident does not run you. When something breaks it opens the incident, names who has it, checks whether a recent release is the likely cause, and follows your written recovery steps. It posts an update on a promised schedule and keeps that promise even when there is nothing new to say. It never changes your production systems on its own.

It writes the postmortem nobody gets round to. A timeline built from evidence, the reasons the system allowed it, and a short list of fixes with an owner and a date on each. Then it chases those fixes, because the ones that never get done are the ones that cause the next outage.

It turns the alarm noise back down. Every alert that fires gets a verdict: somebody acted, it fixed itself, or it should never have fired. The ones that have fired twenty times and produced nothing get proposed for deletion, because an alarm nobody believes is worse than no alarm at all.

Once a month you get one page: what you promised, what you delivered, what the failures cost in minutes, and which agreed fixes are still outstanding.

Works alongside the rest of your product and engineering team.