Nandish Kotadia

Developed a high-performance HTTP server for ads asset serving, handling 5 billion API requests/day using Java Netty. This reduced the response time of the ads serving API by 70% and Redis space consumption by 85%. I optimized Redis operations and implemented the custom Redis bitmap module to efficiently implement the needed Redis set operations (intersect, difference, union). Additionally, I created features to optimize the ads serving system to serve more relevant ads resulting in better CTRs and CVRs, and optimized the ad-click API response time by 50%.

Re-architected the back-end data pipeline for the ads serving platform to handle 100-200 million events traffic daily, providing near real-time reports, which made the system more scalable, performance efficient, reliable, and maintainable. I developed the Kafka stream platform to merge and join events from various Kafka queues generating the joined and aggregated results in real-time-like views, impressions, and clicks. I designed and developed the generic Kafka consumer and producer framework to log aggregated events into MySQL, which can be extended to other databases.