Applied Scientist II, Predictive Targeting

Job :  |  Services LLC

Are you passionate about using data and machine learning to improve how Amazon reaches customers? The Customer Targeting team is looking for an Applied Scientist to build the science behind the targeting of Amazon's marketing initiatives across channels and ad formats. You will turn business questions into modeling problems and take your solutions from prototype to production, working closely with senior scientists, engineers, and business partners.

We are moving targeting toward outcome-based optimization and an AI-native, agentic way of working. In this role you will work on online optimization and reinforcement learning, incrementality modeling, and the next generation of our segment-based targeting model. You will also help build agentic workflows that make targeting decisions transparent and bring science recommendations to the people using them.

The right candidate has strong quantitative and statistical skills, hands-on experience with modern ML methods, and is comfortable working through ambiguity in a fast-paced environment. You pay close attention to detail, learn quickly, and work well in a highly collaborative team.

• Build online optimization (e.g., TSO, bandits) and reinforcement learning methods that optimize targeting decisions for business outcomes rather than proxy metrics.
• Develop incrementality and uplift models that estimate the causal impact of targeting, and use them to improve who we target and how much we spend.
• Improve the existing segment-based targeting model: engineer and evaluate new features, migrate it to new model architectures, and build automated feedback loops that retrain and improve the model from observed outcomes.
• Contribute to the intelligence layer of our agentic system: build components that explain targeting decisions and propose science-driven solutions to business partners.
• Design and analyze experiments (A/B tests, causal inference methods) to measure the incremental impact of targeting changes.
• Write production-quality model code and partner with engineers to deploy, monitor, and maintain models.
• Present methods and results clearly, backed by data, with actionable recommendations for science and business stakeholders.

Predictive Targeting builds the systems that decide who sees Amazon's marketing content, and how often, across Amazon's stores and channels. We are engineers and scientists working on one shared problem: making sure customers see the most relevant and engaging marketing, and making each campaign better than the last.

• 3+ years of building models for business application experience
• Experience in patents or publications at top-tier peer-reviewed conferences or journals
• Experience programming in Java, C++, Python or related language
• Experience in any of the following areas: algorithms and data structures, parsing, numerical optimization, data mining, parallel and distributed computing, high-performance computing
• PhD, or Master's degree and 3+ years of applied research experience

• Experience using Unix/Linux
• Experience in professional software development
• Experience creating processes and improving tools that have measurably improved productivity, quality, or advertiser experience
• PhD in a quantitative field such as statistics, mathematics, data science, business analytics, economics, finance, engineering, or computer science, or experience in defining and creating benchmarks for assessing GenAI model performance
• Experience with reinforcement learning or online optimization in production, such as multi-armed or contextual bandits, exploration vs. exploitation strategies, or real-time pacing and allocation under budget and frequency constraints.

A…