About the job
The Fulfillment by Amazon (FBA) Science team is looking for a passionate, curious, and creative Senior Research Scientist with deep expertise in statistical modeling, machine learning, and large language models (LLMs), and a proven record of solving complex forecasting problems at scale. Our team sits at the intersection of supply chain science, seller behavior modeling, and policy analytics — building the forecasting backbone that powers FBA's shipment creation, inbound arrival planning, and inventory management.
Responsibilities
As a senior member of the FBA Science forecasting team, play an integral role in building and advancing Amazon's FBA shipment creation, inbound arrival, and inventory forecasting systems.
Research and develop statistical models, ML models, and LLM-based solutions to forecast seller shipment creation behavior, inbound arrival patterns, and downstream inventory levels across the FBA network.
Model and quantify the impact of seller behavior changes and FBA policy updates (e.g., capacity limits, fee structures, inbound placement policies) on forecasting accuracy, and develop robust forecasting approaches that adapt to these dynamics.
Build explainability frameworks for forecasting models — enabling science teams, product managers, and business stakeholders to understand model drivers, diagnose forecast errors, and trust model outputs.
Define a long-term science vision and roadmap for the forecasting team, driven fundamentally by customer and seller needs, translating those directions into specific plans for research and applied scientists, as well as engineering and product teams.
Drive and execute forecasting science projects end-to-end: from ideation, analysis, and prototyping through to development, deployment, metrics definition, and monitoring.
Review and audit modeling processes and results for other scientists, both junior and senior.
Advocate the right science solutions to business stakeholders, engineering teams, and executive-level decision makers.
Qualifications
Minimum
PhD, or Master's degree and 5+ years of quantitative field research experience
3+ years of investigating the feasibility of applying scientific principles and concepts to business problems and products experience
Experience with big data technologies such as AWS, Hadoop, Spark, Pig, Hive etc.
Experience communicating qualitative research methods and findings to non-qualitative researchers
Experience in standard machine-learning and statistical modeling tools and techniques (e.g. random forests, gradient-boosted regression, LASSO, logistic regression)
Proficiency working with Python and other high-level languages (Java/C++/Scala) with at least 5 years of coding experience
Preferred
Extensive knowledge and practical experience in several of the following: time-series forecasting, Bayesian methods, causal inference, deep learning, LLMs, and reinforcement learning
Experience building model explainability frameworks for production forecasting systems
Practical experience building and evaluating deep learning models using major frameworks (e.g., PyTorch, TensorFlow, or similar)
Significant peer-reviewed scientific contributions in premier journals and conferences