San Jose, California, United States
As a Computer Science undergraduate at UC Irvine, I am conducting research under Dr. Hayes, specializing in the development of entirely new, non-traditional machine learning (ML) methods. Unlike mainstream AI/ML approaches, this research challenges conventional paradigms and explores innovative techniques that fall outside the standard definitions of artificial intelligence.
During my time working on Amazon's Sponsored Products, I contributed to both the offline and online infrastructure, focusing on feature engineering and optimizing their integration into machine learning models aimed at enhancing ad ranking within the search page. This involved conducting simulations to evaluate the impact of specific features on key performance indicators (KPIs), allowing for data-driven analysis and informed decisions on feature deployment for ranking. To enable real-time availability of these features, I designed and implemented a scalable cache structure using Amazon's Redis, automating the feature onboarding process and streamlining the publishing workflow for increased efficiency.
This year, I had the opportunity to collaborate with the Displacement Cost team within Sponsored Products, where I focused on enhancing their automated DAG (Directed Acyclic Graph) workflows for data collection. I introduced multi-stage validations at critical points of the process, including input validations, intermediate checks, and final step validations. Each validation stage was equipped with real-time monitoring that sent metrics to dashboards and triggered alarms to alert the team of any anomalies, allowing for timely investigation and control over whether the workflow should proceed with data collection. Upon successful completion of the final validation stage, a comprehensive report was generated and securely stored in Amazon S3, ensuring a streamlined and reliable data management process.