At Google, I shape backend infrastructure and observability practices for Google Meet, including resilient AI-powered capabilities used by more than 100M daily participants.
I co-led capacity planning during Meet's 30x pandemic surge, moving toward predictive demand models and a 50x growth target across 20+ global data centers. I also designed flexible cache sharding that increased throughput 1.8x per instance shape and a graceful video-degradation mechanism that protected join success during traffic spikes.
I build production systems that are easier to operate, from one-click capacity activation and binary-rollout workflows to canary-release disciplines and SLO frameworks that reduced deployment-related incidents by over half.
Previously at Amazon, I owned checkout and inventory reliability for Prime Day, helping maintain zero customer-visible checkout outages as events exceeded 100 million products purchased. I built the command-center health dashboard, drove readiness across three Prime Days, and helped establish migration gates and game-day rehearsals adopted by commerce-facing teams.
