At The Trade Desk, I own a real-time bidding pipeline that processes millions of ad decisions per second. I migrated its engine from Python to Go, cutting p99 latency by roughly 30% while maintaining the SLA.
I also partnered with the data platform team to stream ad events into the feature store via Kafka, shortening ML training pipeline lag from hours to under ten minutes. I built shared Python libraries used by roughly 15 services and added dashboards and SLO tracking for the bidding fleet.
At Twitter, I rebuilt the push notification delivery service in Go and shipped a backend for the Tweet engagement metrics API. Earlier at Amazon, I designed backend services for the product catalog and built a Redis caching layer that cut database load by roughly 40% during peak shopping events.
