United States
Led the architecture and launch of Microgenres in 5 countries, enhancing content discovery for 250M+ customers using GenAI (Claude, Llama, OpenAI) tools and prompt engineering and evaluation. Developing and optimizing highly scalable, low-latency distributed microservices in Java and Python, powering carousels on the Prime Video homepage. Data-driven leader & active mentor, fostering technical excellence and team growth.
Designed and implemented a job queuing system for long-running SageMaker jobs, saving 24+ hours of developer time daily for internal and external customers. Architected a PyTorch-based deep learning library to optimize data processing throughput, enabling efficient horizontal scaling across training clusters.
Led the design, implementation, and launch of SageMaker Debugger headless mode in training jobs, a key feature showcased in the re:Invent 2019 keynote. Designed SageMaker Wrangler headless mode and guided its implementation through active mentorship, contributing to the re:Invent 2020 launch. Spearheaded the expansion of the SageMaker training stack to 7 regions.
Willow Ambulatory
Scientific Computation.