Yiyang Pan

About Yiyang Pan

I’m a Software Development Engineer at Amazon specializing in scalable, distributed backend systems that power global payments, financial services, and personalized advertising platforms.At Amazon, I’ve led multiple high-impact initiatives improving system latency, reliability, and targeting precision for large-scale recommendation and merchandising services—ensuring stable ad delivery and performance during global high-traffic events such as Prime Day.Previously, I worked on serverless security analytics and data protection solutions within the Alexa organization, supporting petabyte-scale processing across thousands of AWS accounts. Before Amazon, I contributed to large-scale healthcare data integrations at Oracle Health (formerly Cerner) for the U.S. Department of Defense.I’m passionate about high-performance backend architecture, data integrity, and observability, and I enjoy solving complex problems at the intersection of scale, automation, and cross-service design.Outside work, I love snorkeling, fantasy RPGs, and spending time with my Siberian cat.

Experience

1. Software Development Engineer Ii - Amazon Payment Products

   Amazon

   Jan 2022 — Present · Seattle, WA, US

   Merchandising Automation Program and Learning Engine (MAPLE) is Amazon\'s end-to-end machine learning–driven merchandising platform powering personalized payments and financial services recommendations across the Amazon shopping journey. Delivered key enhancements for Feebas, MAPLE\'s next-generation banner selection service, supporting marketplace expansion and Prime Day with billions of impressions daily while improving latency and targeting precision to ensure reliable delivery under peak traffic. Led latency investigation during load testing, diagnosing cache cold-start issues during fleet scale-up despite multi-layer caching; implemented mitigation strategies (proactive fleet scaling, traffic distribution adjustments, configuration tuning) reducing average latency from 382ms to 160ms and enabling Prime Day launch at 16.7K TPS peak (+4,100% traffic increase). Designed and implemented BrowseNodeSelector for ASIN-level banner targeting in Feebas, improving content precision through cross-service integrations and establishing a scalable framework for future selector enhancements across 20+ marketplaces. Spearheaded cross-functional investigation resolving critical data integrity issues affecting 84 Product Offers ahead of Q4 High Velocity Events. Led technical deep dive with Data Engineering partners to identify the root cause and drove solution implementation across MAPLE systems, ensuring accurate content scoring and impression allocation for high-stakes campaigns. Architected multi-tenant rate limiting across 20+ services to eliminate DynamoDB throttling in IssuanceAdminService, preventing cascading failures to multiple Tier-1 services and ensuring reliable content delivery during high-traffic periods like Prime Day. Delivered four globally adopted features in MAPLE\'s advertising content management platform, streamlining campaign workflows, eliminating performance bottlenecks, and saving over hours annually for global marketing teams.

Education

• Purdue University

  BS, Computer Science