Costs were climbing faster than usage and nobody owned the number, so every team assumed someone else was watching it. I took the problem, instrumented spend down to the service level, and worked through the biggest offenders one at a time until the trend line bent and stayed bent.

Decomposed the monolith behind a strangler proxy into 4 services — no maintenance window needed.

The pipeline lost rows quietly under load and we only found out when a customer's numbers didn't add up, which is the worst possible way to find out. I chased it to the root — backpressure, at-least-once handling, and reconciliation that proved nothing fell on the floor rather than hoping it hadn't.

Moved data and traffic closer to users; cross-region tail latency fell 35% with no consistency regressions.

The system was buckling under growth and the old design had run out of room — every fix bought weeks rather than quarters, and the next step up in traffic would have broken it outright. I owned the redesign end to end: the tradeoffs, the migration path, and a rollout that kept old and new live until the cutover was boring.

Right-sized clusters and added autoscaling; the monthly bill dropped 20% at the same SLOs.