Software Development Engineer

The Off-Search Online Sourcing team builds the real-time ad retrieval systems behind every Detail Page, Home Page, and non-Search surface on Amazon — searching hundreds of millions of ads in milliseconds using embedding models and approximate nearest neighbor indexes. You'll design and deliver the low-latency services, model serving infrastructure, and distributed indexing pipelines that determine which ads shoppers see, enriched with delivery speed, locality, and return signals so ads surface products that can reach customers quickly. This is a team that ships fast — rapidly experimenting with new models, indexes, and signals in production through A/B experiments to grow both advertiser results and shopper engagement. You'll partner closely with applied scientists to bring new retrieval models from research into production, and use experimental results to decide what ships. Key job responsibilities - Design and develop high-throughput, low-latency services that retrieve and enrich ad candidates within strict latency budgets across Amazon's non-Search surfaces - Build and scale model serving infrastructure for real-time embedding inference, partnering with applied scientists to bring new retrieval models into production - Develop distributed indexing pipelines that build and refresh large approximate nearest neighbor search indexes across hundreds of millions of ads - Deliver real-time fulfillment signals — delivery speed, locality, return eligibility — that improve ad filtering and ranking quality for shoppers - Launch new retrieval models and indexes behind A/B experiments, analyze results, and use data to decide what ships to production - Deliver features end to end — from design and code reviews through implementation, deployment, monitoring, and on-call operations - Improve operational excellence through caching, graceful degradation, monitoring, and root-cause analysis - Mentor peers through code and design reviews, helping raise the engineering quality bar across the team A day in the life On a typical day you might start by reviewing latency dashboards for your retrieval services, then dive into building a new distributed pipeline to refresh a nearest neighbor index. After lunch, you could be partnering with an applied scientist to integrate a new embedding model into the serving path, setting up the A/B experiment framework to measure its impact. Between coding sessions, you'll review pull requests, debug a latency regression, and check on the results of a running experiment to decide whether a new signal improves shopper engagement enough to ship. About the team The Sponsored Products and Brands team at Amazon Ads is re-imagining the advertising landscape through industry-leading generative AI technologies, revolutionizing how millions of customers discover products and engage with brands across Amazon.com and beyond. We are at the forefront of re-inventing advertising experiences, bridging human creativity with artificial intelligence to transform every aspect of the advertising lifecycle from ad creation and optimization to performance analysis and customer insights. We are a passionate group of innovators dedicated to developing responsible and intelligent AI technologies that balance the needs of advertisers, enhance the shopping experience, and strengthen the marketplace. The Off-Search Online Sourcing team within SPB builds the real-time systems that choose which ads shoppers see on the Detail Page, Home Page, and other Amazon pages — searching hundreds of millions of ads in milliseconds using embedding models and approximate nearest neighbor indexes, and enriching every candidate with delivery speed, locality, and return signals so shoppers see ads for products that can reach them quickly.