Harut Avetisyan

At Uber, I build internal capacity-platform products from customer conversations through backend, data, frontend, quality, and alerting delivery. I'm a core contributor to the Platform Efficiency Portal, which gives 81+ platform engineers and leaders real-time visibility into $100M+ in infrastructure allocation and cost.

I own the ingestion layer behind Uber's infrastructure cost estimator and built dynamic-SQL services over Pinot with 1B+ records and sub-400ms queries. My pipelines ingest 120M+ metrics daily into Hive and Pinot, turning GPU, capacity, and infrastructure telemetry into usable dashboards.

I also built data-quality and reliability systems using IQR, Z-score, Statsmodels, M3DB, and uMonitor, reducing incident response from about an hour to 10–20 minutes. I established integration-testing and incident-management practices that helped the team reach Uber's top-three company-wide Deployment Safety Score.

Previously, I built RAG message-personalization pipelines at Imea Systems, improved engagement by roughly 35%, and helped migrate a monolith to an event-driven architecture. Across Sentium Consulting and EKey, I delivered ETL pipelines, accelerated MinIO uploads by about 85%, and decomposed monoliths into scalable asynchronous services.