Apr 2025 – Present · 11 mos

• Google Lens & Search Ranking.
• Technical Lead for low-latency and low-capacity query understanding system to map visual queries to semantic ”bag-of-words” signals; this ML technology improved ranking of results for high-volume Google Lens and Multisearch queries, driving increase in relevant clicks and user growth.
• Architected a strategic roadmap to enhance query understanding system quality by identifying critical loss patterns and engineering novel success metrics with scalable ML tuning frameworks; this helped drive widespread cross-product adoption of the improved signal, directly unlocking revenue improvements for Lens Ads and powering features across Google Images, Lens Web, and AI Overviews.
• Enabled Lens to understand dense-text query images by engineering a novel low-latency OCR relevance model trained on LLM generated data using a parallel data-processing pipeline

Oct 2022 – Present · 3 yrs 5 mos

Mar 2021 – Oct 2022 · 1 yr 7 mos

• Cloud Spanner Database team.
• Architected critical observability features for a large-scale distributed database; exposing high-cardinality metrics and stats (latency distributions, ops-per-table, cpu-usage) to enable customers to self-diagnose complex performance bottlenecks like long tail latencies, lock contentions, table hotspotting and high CPU usage problems; reduced customer escalations by 2x.
• Built a high-throughput OpenTelemetry observability pipeline to process, export and visualize Google Cloud statistical-data as metrics in the customers' APM tool of choice.