Software Development Engineer, EC2 Capacity Telemetry and Autonomous Systems

The EC2 Capacity Telemetry and Autonomous Systems team is looking for an experienced Software Development Engineer to join our Seattle-based team. EC2 (Elastic Compute Cloud) is part of Amazon Web Services, the global leader in public cloud infrastructure.

We build the systems that measure what EC2"s capacity is doing in real time - what exists, what is in use, what is reserved, and where demand is going unmet - detect when any of it moves, establish why, and increasingly act on it without a person in the loop.

We sit within a group of software engineers and scientists who build the forecasting algorithms and modeling systems that determine how EC2 is built and packed, so that capacity is maximally utilized and maximally available through automated capacity planning and optimization. Those models run on what our systems measure, and the decisions they produce are executed by software we own.

This is a software engineering team at the crossroads of writing our own systems and operating the best of open source. We design and run distributed services with real availability and latency commitments. We build low-latency query and serving layers on engines such as Apache Druid, ClickHouse and Apache Pinot, and extend them where our workload outgrows what they do out of the box. We write optimization and decision engines that replace work people were doing by hand. We build APIs that other services depend on, and the React web applications that operators and planners run their day from. And we build agentic systems on top of all of it, so a problem gets diagnosed and resolved by software rather than escalated to a human with a chart attached.

The work is operationally real. Our consumers are live systems, and ours matter most at exactly the moments something else is going wrong, so availability is a design constraint here rather than something handled after launch.

Key job responsibilities

• Design, build, launch and operate production services end to end - you will write the design document, make the trade-offs, defend them in review, and own what you shipped
• Build high-throughput distributed systems that operate under real latency and availability commitments, and the query and serving layers behind them
• Build real-time detection and alerting that tells an operator what changed and why it matters, rather than requiring them to know what to look for first
• Build rules and decision engines that replace large-scale manual workflows with software
• Build optimization systems that return the best available answer within a fixed time budget, degrading to a simpler strategy rather than failing
• Build and extend the APIs other services depend on, and the React web applications our operators and planners work in
• Build agentic systems that investigate a problem, determine the cause, recommend a course of action and increasingly take it
• Raise the operational bar on what you own: monitoring, alarming, recovery paths and the tests that span service boundaries

A day in the life

Your day is a mix of building and deciding. A typical week has you deep in implementation on a service you own, in a design review arguing about someone else"s, and in a conversation with the people who depend on what you build.

Your closest collaborators are the engineers and scientists who build the forecasting and optimization models next to you - they need what your systems produce, and what they produce has to land in systems you operate. Beyond the immediate team you will work with capacity planners and operators who make decisions from the software you build, with partner service teams whose systems you integrate with and occasionally contribute code to directly, and with the leadership who review the operational numbers your systems generate.

You will be expected to take on ambiguous problems and work independently, making business and technical trade-offs to deliver results - with support from teammates who will answer questions and work through day-to-day problems with you. Where no established pattern exists, we evaluate options, choose one, and write down why.

You will be a fully participating team member in daily stand-ups, design review sessions, demos, code reviews and other team gatherings. This position involves on-call responsibilities, typically one week every two months. We build fault-tolerant systems to minimize being paged during off-hours, and when we are paged we root-cause and fix so the same page does not come back.

About the team

Our mission is that EC2 capacity is maximally available and maximally utilized, managed autonomously rather than by hand. Every system we build moves work from a person to software and then makes that software accountable for the outcome.

We set tenets and target outcomes before a design exists, then build against them and settle the trade-offs in review. Design, implementation, prototyping, security review and documentation rotate across the team, so no system has a single point of knowledge and engineers end a year with more range than they began it.

We are invested in your growth. We assign projects that balance what the business needs against what will make each engineer more capable of owning something larger next time. As an experienced engineer you will have opportunities to mentor while working alongside senior and principal engineers across the organization. As a team, we enjoy connecting socially while respecting the balance between work and personal life.

As an experienced engineer you will have opportunities to mentor while working alongside senior and principal engineers across the organization.

As a team, we enjoy connecting socially while respecting the balance between work and personal life.