Software Development Engineer, Marketing Measurement and Performance Science

Marketing Measurement and Performance Science (MAPS) measures the incremental impact of Amazon's marketing investments on customer perception, action, and purchases — covering billions in annual fixed marketing (FM) spend across the full funnel. Our outputs provide spend insights and recommendations that power total investment decisions at the S-team level for OP planning, BU-level budget allocation, and in-year spend guidance. As an SDE II, you'll design and build the systems that transform and harmonize data from 20+ external partners — each with their own schemas, grains, cadences, and nuances — into the measurement-grade inputs that power COSMOS, Amazon's FM causal measurement framework. The problems you will coverage go beyond standard ETL responsibilities — you'll build AI-native systems that reconcile providers with incompatible definitions, handle split metric ownership, manage retroactive revisions, and maintain the immutable snapshots and deterministic joins that causal inference demands. Key job responsibilities - Design, develop, and maintain measurement-grade data systems at scale — ingesting, standardizing, and vending marketing data from 20+ external and internal sources that feed causal modeling and MLOps systems. - Own full lifecycle delivery of production software on complex, ambiguous problems — from design through launch and ongoing operations — with independence and minimal guidance. - Build automated validation pipelines and data quality frameworks that enforce contracts, detect anomalies across providers, and ensure measurement-grade integrity at every stage. - Define statistical methods for outlier detection, diagnose root causes systematically, and determine corrective actions to maintain data trust. - Partner across science, product, and engineering teams to scope solutions, navigate constraints, and ship the most efficient path from prototype to production. - Write clean, well-tested code (Python, Scala, or Java) and mentor junior engineers on system design, code quality, and operational best practices. A day in the life Day to day, you'll build and scale multi-layer automated validation pipelines, with clear data lineage so every model run is fully reproducible. Our vision is to scale our infrastructure across new business units and geographies reaching 90%+ coverage of Amazon's FM spend, and develop self-service catalog and observability tooling that lets scientists and partner teams explore our data without filing tickets. You'll also have a direct influence on schema governance — designing systems that enforce data standards at the point of contract, detect drift from providers, and keep our specifications current as partnerships expand. You'll collaborate closely with causal scientists, economists, product managers, and agency data ops teams — translating measurement requirements into scalable technical solutions. This is a high-ownership role where your work directly determines whether Amazon's leadership can trust the numbers behind billion-dollar marketing investment decisions.

Within MAPS, the Marketing Inputs & Data Automation (MIDA) team owns the measurement-grade data layer that sets the ceiling on what our causal models can measure, where they can operate, and how confident leadership should be in the outputs. We build and operate large-scale data infrastructure and data assets — ingesting, validating, harmonizing, and vending data from 20+ third-party providers (agencies, aggregators, publishers) across multiple Amazon business units and marketing channels, with global coverage. Our pipeline is purpose-built for the high bar of causal inference — not dashboards or reporting — requiring strict temporal integrity, historical stability, multi-layer validation, full lineage, and reproducibility at every stage.