Production support lead
Running the platform after everyone else has gone home, and making the next incident less likely than the last.
I lead L2 production support for boeingdistribution.com, a B2B commerce platform where downtime stops parts reaching aircraft. The job is incident response, root-cause analysis and the slower work of building observability so the same failure does not recur — thread and heap dumps, AppDynamics and Splunk, JVM and cluster tuning, and release coordination that makes deployment boring.
Incidents are a measurement problem first
Most production arguments are about what is wrong, not how to fix it. Support blames the database, engineering blames traffic, and nobody has evidence. The first job is almost always instrumentation, because a fix applied to the wrong cause is worse than no fix — it consumes the outage window and teaches the team the wrong lesson.
Migrations without an outage
I have run a data centre migration on a live storefront with minimal disruption, and kept a high-traffic site up through sustained growth via JBoss tuning and cluster expansion. Neither is glamorous work and both are the difference between a platform people trust and one they route around.
Mentoring rather than escalating
A support team that escalates everything never gets stronger. Most of what I do day to day is walking engineers to the root cause themselves, which is slower once and faster permanently.