Description

Amazon is investing heavily in building a world-class advertising business, and we are responsible for defining and delivering a collection of advertising tools and products that drive discovery and Advertiser success. Our products are strategically important to our Retail and Marketplace businesses, driving long-term growth. We deliver billions of ad impressions and millions of clicks daily and are breaking fresh ground to create world-class products. We are highly motivated, collaborative, and fun-loving with an entrepreneurial spirit and bias for action.

The Marketing Effectiveness & Attribution Science team develops causal inference and machine learning systems to measure the impact of marketing programs across Amazon's advertising ecosystem. We build production-grade attribution models that help business teams understand what's working, optimize resource allocation, and drive advertiser growth. Our work sits at the intersection of econometrics, scalable ML systems, and high-stakes business decisions.

As a Data Scientist on this team, you will own end-to-end modeling pipelines — from problem formulation and experimental design to model development, productionization, and stakeholder communication.

Major responsibilities include:

Translate / Interpret:

Partner with cross-functional teams to translate business questions into rigorous causal inference problems
Design observational studies and quasi-experiments to measure marketing effectiveness when traditional A/B tests are infeasible
Work with data engineering to instrument new data pipelines when existing data cannot answer the causal question

Measure / Quantify / Expand:

Own and evolve production attribution models across multiple marketing channels
Build and maintain causal inference pipelines using methods such as Difference-in-Differences, Synthetic Control, Double Machine Learning, and Media Mix Models
Develop scalable PySpark and Python codebases that process large-scale event data
Continuously improve model accuracy through feature engineering, heterogeneity analysis, and sensitivity testing

Explore / Enlighten:

Investigate anomalies in model outputs and deep-dive to identify root causes
Develop automated data quality checks and model diagnostics
Research and prototype next-generation measurement methods

Make Decisions / Recommendations:

Present findings to senior leadership with clear recommendations
Build dashboards and self-service tools that enable stakeholders to explore results independently
Write production-quality Python code for data analysis, model training, and result publishing