Mountain View, California, United States
BigQuery Team
Technical Lead for the architecture and redesign of the next-generation Dataproc Metastore: a fully managed, highly available, multi-regional, auto-healing serverless metastore supporting Apache Hadoop ecosystem workloads.
• Re-architected both control and data planes from GKE to Cloud Run, moving our service to a serverless deployment.
• Reduced infrastructure costs by 72% and improved startup latency by 55% while enhancing scalability and reliability.
• Ensured full feature parity between legacy and serverless systems, adapting 8 years of prior feature work to new, serverless acrchitecture, enabling seamless transitions for customers.
• Designed and executed zero-downtime migrations, with no impact to production workloads.
• Cut inactionable oncall alerts by 88%, improving operational efficiency and team productivity.
BigLake Metastore & Iceberg REST Catalog
Contributed to the development of BigLake Metastore: a unified, serverless metadata platform that connects Cloud Storage and BigQuery data to engines like Apache Spark, Apache Flink, and BigQuery SQL.
• Delivered features enabling cross-engine metadata interoperability via the Apache Iceberg REST catalog.
• Helped establish a single source of truth for Iceberg data, improving consistency and discoverability across compute engines.