•Trained ML ranking model for Amazon Music voice search personalization - Developed features incorporating user likes signals (tracks,
albums, artists, playlists) across 200M+ customers, improving search relevance for both recognized and unrecognized user profiles.
•Architected scalable feature engineering pipeline for ML ranking system - Evaluated 5 distributed system approaches (Bloom filters,
search index federation, direct API calls); recommended solution reducing costs by $371K/month while maintaining sub-100ms p99
latency at 2.2K TPS, enabling reuse across 38+ client applications
•Designed and developed a GenAI-based Code Reviewer leveraging LLMs to analyze source code, detect bugs, and recommend
performance and readability improvements — reducing manual review time by 60%.
•Led team-level peak readiness for Amazon Peak Events (Prime Day, Black Friday/Cyber Monday) - Partnered with senior
program managers to drive peak related activities and working backwards dates for code freezes, executed load and chaos testing
across team owned services, validated projected peak volumes for capacity planning, and scaled services accordingly, enabling the
backend stack to handle ~100K TPS during events.
•Mentored 2 new-grad engineers and 3 interns, providing technical guidance, code reviews, and career development support; fostered a
culture of excellence through knowledge-sharing sessions and coding best practices.