AWS Insights & Optimization is looking for an Applied Scientist to help develop sophisticated algorithms and models that involve analyzing and learning from over 540 billion customer cost, usage, and utilization events daily. We use this data to power cost anomaly detection, cost forecasting, rightsizing recommendations across seven compute services, Savings Plans optimization, and AI-powered conversational experiences that help customers understand and optimize their AWS spend. Our team's vision is to be the world's provider of intelligent AWS cloud financial management, where customers can understand, control, and optimize usage of AWS products. We sit at the nexus of all AWS services and interact directly with end-customers, building relationships with teams across AWS to ensure we offer a secure and reliable experience that builds trust and provides intelligent insights. As a successful Applied Scientist in AWS Insights & Optimization, you will own models end-to-end — from problem formulation through experimentation to production deployment. Your work may span cost anomaly detection (decomposition, detection, and root cause attribution), time-series forecasting with a focus on accuracy and consistency, rightsizing engines for EC2, EBS, Lambda, ECS, RDS, and Aurora, LLM evaluation science for AI-powered agent experiences, or agent memory architectures that enable persistent, adaptive behavior across sessions. You will work closely with applied scientists, software engineers, and product teams to enhance existing models and build new ones that solve challenging customer problems. You will drive implementation of proposed models, establish testing strategies to validate them before and after production, and define evaluation metrics that determine whether capabilities meet the quality bar. We value accuracy over speed, measurability over intuition, and simplicity over complexity — the simplest model that meets the bar wins. You are an analytical problem solver who enjoys diving into data, are excited about investigating and developing algorithms, and can influence technical teams and business stakeholders to solve real-world customer problems.