Mar 2025 – Present · 1 yr 2 mos · On-site

• Core Data Infra - Building planet scale High Throughput, Low Latency Indexing system that powers Google Search, Ads, Youtube and many other product areas within Google.

Jan 2022 – Feb 2025 · 3 yrs 1 mo · On-site

• Datastream - Datastream is a GCP Service for streaming data from cloud, on-premises, and SaaS data sources to data destinations such as GCS and BigQuery. Data is meant to be streamed in order to be read and processed by multiple source-agnostic readers.
• Architected the modernization of the Oracle CDC (Change Data Capture) mechanism to enhance data streaming reliability by ensuring only committed data is processed.
• Delivered a novel, ROWID-based concurrent backfill solution for Oracle, which:
• Eliminated a 100-million-row limitation, enabling the backfill of multi-petabyte tables.
• Enabled backfill for tables lacking a primary key or index, broadening the feature's applicability.
• Improved customer experience and reliability by mitigating issues caused by inefficient query plans.
• Optimized Oracle source backfill performance and developed a comprehensive performance benchmarking and automation suite for both CDC and backfill, enabling proactive identification of performance bottlenecks.
• Designed and implemented an optimal concurrent batching strategy for parallel backfills, further improving throughput and efficiency.