Notes / 2025
What AT&T and Mozzartbet Taught Me About Systems
Nobody puts "answered tickets for five years" on a highlight reel. It doesn't sound like a career, it sounds like a waiting room. But those years are where I actually learned how systems fail, and nothing since has taught me faster.
It started at AT&T, on a support desk in Brno, taking calls from people who didn't care how the system worked, only that it didn't. That's the part junior engineers skip past too fast. You don't understand a system by reading the architecture diagram. You understand it by hearing the same complaint from forty different people and realizing the diagram left something out. Every recurring ticket is a bug report on the design, whether anyone labeled it that way or not.
Mozzartbet came next, and the stakes changed shape. This wasn't "my file sync is slow," this was a live betting platform where a few minutes of downtime meant real money and real angry customers, in real time, during a match that wasn't going to pause for anyone. Sysadmin work there was less about theory and more about instinct built from repetition. Which logs to check first. Which service usually fails before the others. What a bad deploy looks like in the first ninety seconds versus what a normal traffic spike looks like. You don't get that instinct from a course. You get it from being the person who has to fix it while people are watching the scoreboard.
The unglamorous part is also the honest part. Support work strips away the fantasy that engineering is mostly clever solutions. Mostly it's someone's password reset, someone's misconfigured router, someone's account stuck in a state nobody documented. You either find that tedious and leave, or you start noticing the pattern underneath the tedium, and that pattern is the actual job. Every system I've built since, from pipelines to backend services, carries fingerprints from those two jobs. I default to asking what breaks first, not what looks impressive in a demo.
Five years answering tickets didn't feel like career progress at the time. It felt like standing still while other people wrote code. Looking back, it was the best training I never planned for, and I'd tell anyone starting out in engineering to spend real time on a support desk before they touch anything with a deploy button. You'll ship worse demos and build better systems.