• Building data infra management products for engineers, data scientists, and analysts at FB scale. Leading a team of software engineers with product development, execution and delivery.
• Architected, spearheaded and launched a FB-wide metric platform (Metric360), a single source of truth for company metrics, resulting in 2.5 mil QPM to be executed in heterogeneous compute engines.
• Designed the platform to be privacy-aware and the context is carried throughout the infra stack
• Reduced 100k+ duplicate metrics using metric correlation (Pearson Coefficient, Euclidean and Time Warping) that resulted in improved operational workload for engineers.
• Worked on Facebook Ads Data Infra and redesigned the data flow of Ads Clicks/View Through Conversion; optimized the service so both impression and click events are logged and joined at scale that gave more visibility into Ads performances.
• Conceptualized and drove column level data lineage for the entire warehouse cluster. This resulted in out-of-the-box impact analysis (dependency graph between code and data) that helped in reducing the production error rate and improving dev efficiency.
• Reduce the Ads Data Infra compute weight by optimizing the data flow.
