Ads Infra | AdMonitor

Designed and drove end-to-end rearchitecture — of an near real time ad revenue stage based processing framework, Meta's source of truth for ads revenue, achieving 40% latency improvement (>2.5 sec reduction) and +25% scalability for high-traffic accounts processing 50K+ events/sec/shard of 3000 shard system.

Architected core framework components — Credit Limit Updater, PolicyZone-aware batching, PoolExecutor/ShardExecutor patterns, and rollback-on-restart capability — enabling zero-downtime migration of 14 production stages handling billions of daily ad events

Unlocked $130M+ (0.108% iRev) annual revenue opportunity — through latency improvements enabling better ad pacing accuracy, exceeding latency targets by 5x while unblocking new optimization opportunities (IBT/HBT) for H2'25

Built reusable infrastructure adopted across Ads org — generalized skip functionality, configurable per-stage settings, and parallel-to-pool conversion patterns — saving 11+ engineering-years over 5 years in reduced complexity

●​ Led architecture and development of real-time data streaming systems using Apache Kafka for large-scale media data pipelines.

●​ Reduced end-to-end data latency by 30%, improving timeliness and accuracy of media encoding and distribution workflows.

●​ Drove system design decisions and mentored engineers while scaling distributed systems supporting TV and radio data processing.

●​ Improved system reliability and throughput through better partitioning, backpressure handling, and pipeline optimization.

Working as a lead developer of Clinical Trials Reporting Program (CTRP) 5.0, a comprehensive database of National Cancer Institute-supported clinical trials

Using Rails MVC framework to serve data for front-end apps through REST calls.

Worked on front end web application using Angular JS framework.