I build high-load .NET systems.

The service buckled at peak load. I reproduced it under representative traffic, profiled down to the root cause, tuned the configuration, and validated the fix before promoting it to production. One of those problems where the change is small but finding it is the whole job — and the payoff is users never noticing the thing that used to break.

On-call triage was slow because the signal was spread across dashboards and logs. Nobody was assigned to fix that, so I proposed a small CLI, built it, and drove its adoption across the team. It pulls the picture together into one structured summary — a rough edge smoothed for everyone who carries the pager.

Services needed to react to each other’s state changes without brittle coupling. I designed event-driven messaging with an auto-retry dead-letter path, so a transient failure retries itself instead of paging someone at 2am. The design was clean enough that other teams on the platform picked it up.