Software Development Engineer III, EC2 Core Provisioning

EC2 is the foundation of AWS, and our team sits at the heart of it: we own the systems that take raw server hardware and turn it into sellable EC2 capacity. Every host in the EC2 fleet — across every region, from Graviton to GPU platforms — flows through our provisioning, vetting, and repair workflows before customers can launch instances on it. We are looking for a Software Development Engineer to drive the next generation of our fleet ingestion and dwell-reduction charter. You will own the architecture of the services, workflows, and analytics that shrink that dwell: large-scale orchestration systems, event-driven reconciliation services, and the data pipelines and dashboards that give engineers and leadership real-time visibility into fleet health. As a senior engineer, you will operate with significant autonomy and ambiguity. You will decompose multi-quarter problems into deliverable milestones, influence roadmaps across partner teams (hardware engineering, datacenter operations, capacity planning), raise the operational and engineering bar for the team, and mentor engineers at all levels. You will be a force multiplier: the designs you write, the standards you set, and the engineers you grow will matter as much as the code you ship. Key job responsibilities - Own the end-to-end architecture of high-scale distributed systems that provision, vet, and recover EC2 hosts across all AWS regions - Lead the design and delivery of the team's dwell-reduction initiatives, from data ingestion and analytics through automated remediation workflows - Author design documents and drive technical decisions that span multiple teams and services; influence partner-team roadmaps - Raise the bar on operational excellence: drive down operational load, improve observability, and lead root-cause analysis for complex fleet-level issues - Mentor and grow SDE Is and IIs through design reviews, code reviews, and hands-on guidance - Work directly with product, hardware, and datacenter stakeholders to translate ambiguous business problems into technical strategy A day in the life You might start the morning reviewing a design for a new reconciliation workflow, spend midday deep in a fleet-wide dwell regression — correlating provisioning telemetry across regions to find the systemic cause — and end the day pairing with a junior engineer on a replay-safe workflow change. You'll balance long-term architecture with the operational reality of running Tier-1 services that the entire EC2 fleet depends on.