Aug 2024 – Present · 1 yr 9 mos · Bengaluru, Karnataka, India · Hybrid

• Re-architected a critical metadata API that was repeatedly breaching latency SLAs by removing redundant database calls and heavy serialization; cut P99 latency from ~10s to 400ms, eliminating pager noise and reducing on-call burden.
• Optimized a hot-path database query that caused full table scans in a high-traffic service; reduced P99 latency from >10s to ~760ms and lowered database CPU from 55% to <10%, improving service reliability and scalability.
• Designed and productionized a fault-tolerant, high-throughput retention workflow to purge stale lineage at massive scale (600M+ records) while protecting database replication and service health; sustained cleanup of 25M+ records/day with zero replication lag and no system degradation.
• Built a real-time streaming pipeline using Flink to capture dataset change events and publish standardized lineage events; increased end-to-end lineage coverage from 80% to 96%, enabling safer dependency analysis.
• Improved data asset ownership assignment (used for cost attribution and support routing) by making rule evaluation more precise and leveraging upstream pipeline signals; increased correct ownership coverage from 65% to 90%, reducing misrouted tickets and speeding incident resolution.
• Led a large-scale ownership cleanup across two orgs by reassigning ~55,000 data assets from broad org-level owners to the right team owners using metadata and lineage; improved accountability and day-to-day operational efficiency.