Software Development Engineer, Ads Analytics & Insights-Fixed

About the role

Amazon Ads works with hundreds of thousands of advertisers, each on a different journey — some running their first campaign, others expanding into streaming TV or display, some at risk of slowing down. The Sales AI team builds the machine learning products that tell sales teams where each advertiser is on that journey and what will help them grow. Our segmentation platform models how advertisers mature over time, predicts where each is heading, and identifies the actions most likely to move them forward — shaping how Amazon Ads organizes sales coverage and decides which advertisers receive managed support.

We're now evolving segmentation from describing where advertisers are to recommending what sellers should do: personalized, interpretable recommendations delivered directly in the tools sellers already use, with a feedback loop that measures which actions actually drive advertiser growth. You'll design and deliver the ML platform and data pipelines that take models from experimentation to production, deliver predictions and recommendations reliably to sales systems at scale, and help move from monthly batch outputs toward on-demand insights — working closely with applied scientists, data scientists, and product partners at the intersection of ML, large-scale data systems, and sales productivity.

Key job responsibilities

- Design, develop, and operate the ML platform that takes segmentation models from experimentation to production — including feature ingestion, training, batch inference, and model monitoring

- Deliver model outputs — predictions, scores, and recommendations — reliably to downstream sales systems and partner teams, with clear freshness, versioning, and availability guarantees

- Build scalable, automated data pipelines that bring new advertiser and sales engagement signals into the models quickly, with strong data quality checks

- Design and deliver solutions for well-scoped problems where the technical approach isn't yet defined — such as serving predictions on demand or generating explanations for recommendations

- Partner with applied scientists and data scientists to productionize new models, shorten iteration cycles, and make it easy for scientists to ship without heavy engineering support

- Build the instrumentation that captures how sellers use recommendations and what happens next, so the team can measure impact and improve the models

- Deliver your services end to end — testing, deployment, monitoring, operational excellence, and on-call

- Participate in design and code reviews, holding a high bar for quality, security, and reliability

- Mentor teammates and help onboard new engineers

A day in the life

Your morning might start with the scientists on your team, reviewing results from a new model experiment and deciding what it takes to get it into production. In the afternoon, you might design a pipeline for new advertiser signals, review a teammate's code, or agree with a partner team on how they'll consume a recommendation dataset. Some days you'll trace a data quality alert to its source and fix it; others you'll prototype on-demand predictions. You'll spend most of your day hands-on in the code, and see your work shape how sellers engage with advertisers.