Are you passionate about translating scientific research into real-world impact at massive scale? PXT Central Science (PXTCS) builds the analytical foundations that shape how Amazon attracts, develops, and retains talent at a scale no other company operates. Our models and research directly influence decisions affecting over a million employees worldwide, from compensation and attrition strategy to employee engagement and workforce planning
The L5 Research Scientist in Survey Science contributes to the science behind Amazon's state-of-the-art employee surveys, setting the standard for how Amazon measures what its employees are experiencing. The questions that matter most in this role are the ones numbers alone cannot settle, and that often run ahead of demand — identifying the challenges leaders will face before they ask, and building the methods to address them. How is the culture changing as the company evolves? What are employees experiencing that no metric on a dashboard captures? Where are employee concerns emerging that have not yet shown up in any data? These factors drive business outcomes, resist clean quantification, and require creative measurement approaches to address.
Key job responsibilities
- Design, develop, and evaluate mathematical, statistical, and machine learning models to address complex business problems, ensuring solutions are scalable, reliable, and performant in production environments.
- Apply a breadth of scientific techniques — including generative AI approaches where appropriate — to frame ambiguous business challenges into well-defined research problems, working backwards from customer needs to identify the right methodology.
- Collaborate with software development engineers and cross-functional partners to translate research prototypes into engineering specifications and production-ready systems, balancing scientific rigor with practical constraints.
- Provide guidance on survey methods to teams across PXT, so employee research across Amazon meets a consistent standard.
- Develop and audit metrics to quantify solution impact, measure project progress, and gate rollouts based on quality, while identifying opportunities for continuous model improvement.
- Mentor other scientists and contribute to the broader scientific community through peer reviews, publications at internal or external venues, and active participation in hiring and team planning.
A day in the life
You might start your morning reviewing experimental results from a model you deployed last week, then shift into a design discussion with engineering partners about integrating a new approach into an existing system. After lunch, you could be deep in data analysis, exploring patterns that inform your next hypothesis. You'll regularly present findings to teammates, provide scientific reviews for peers across the organization, and carve out time to stay current with the latest research in your field.
About the team
PXT Central Science is Amazon's people-science R&D organization. We develop causal inference methods, predictive models, and decision-support tools that operate at the intersection of economics, machine learning, and organizational science. Our work spans employee experience, workforce planning, compensation, talent acquisition, and employee relations — always grounded in rigorous methodology and always aimed at real-world impact.
