Amazon has the world's most complex supply chain: we fulfill global demand for hundreds of millions of products at lightning fast delivery speeds. We need your skills to optimize our supply chain, with the end goal of delighting our customers. A core part of the supply chain operations is Demand Forecasting: We forecast the demand of tens of millions of products. These forecasts are used to make many decisions, such as automatically order hundreds of millions worth of inventory, decide where to place that inventory, and establish labor plans for hundreds of warehouses. The Product Data Science team within Forecasting & Labs (part of Supply Chain Optimization Technologies) is looking for an analytical and technically skilled Data Scientist to join our team. Our team is responsible for bias correction model development and A/B testing for forecast improvements, GenAI/LLM research for forecast explainability, and deep analytics for Labs and Foundation Models. We work horizontally across the forecasting product portfolio—including National, Regional, Grocery, SSD, Inbound, and CIV forecasting—to embed advanced analytics and machine learning solutions where they create the most value.

Design and analyze experiments (A/B tests) to measure the impact of forecast model changes and SCOT initiatives, drawing causal inferences from both experimental and observational data

Develop bias correction models to improve forecast accuracy across Amazon's demand forecasting systems, including National, Regional, Grocery, SSD, Inbound, and CIV forecasts

Contribute to GenAI/LLM-based research for forecast explainability and interpretability, helping stakeholders understand what drives forecast signals

Support and enhance the Labs experimentation platform by building scalable inference and measurement solutions that quantify the impact of forecasting improvements

Work horizontally across the forecasting product portfolio and collaborate with product managers, applied scientists, and engineering teams to embed analytics and ML solutions where they create the most value

Use large datasets to build models addressing ambiguous forecasting questions, including demand prediction, out of stock, seasonality, and varying lead times and spans

Interpret data, write reports, and communicate measurement results to stakeholders by translating technical frameworks into business-oriented insights and actionable recommendations