Software Development Engineer III, AWS AI/ML Servers

AWS runs the world"s largest fleet of AI/ML accelerator servers. When a model with billions of parameters trains across a large scale of GPUs, the software that provisions, monitors, and heals that fleet is what keeps training on track. We build the services, control planes, and data platforms that turn hundreds of thousands of servers into a reliable, self-managing system. If you want to build software at the intersection of large-scale distributed systems and the physical infrastructure powering frontier AI, this is the role.

We are seeking a Software Development Engineer to design and build the software that operates our accelerated compute fleet: fleet-management services, orchestration and remediation control planes, telemetry and analytics pipelines, and the APIs and tools that engineering, operations, and partner teams depend on. You will own software from design through production, working across multiple teams and organizations to deliver scalable, reliable services.

What You Will Do

You will design and build distributed services and data platforms that manage the fleet end to end. You will turn ambiguous problems into well-architected software: designing APIs, choosing data models, making the latency, availability, and cost trade-offs that come with operating at fleet scale. You will build the control planes that automate provisioning, health evaluation, and remediation, and the pipelines that transform raw telemetry into signals other teams act on. When your service catches and routes a failing node automatically before it disrupts a training job, that is your impact.

Why You Will Love It

Your software runs across a large scale of servers in the cloud, and you see its impact on fleet availability within days of shipping. You solve genuinely hard distributed-systems problems, own services end to end, and work close enough to the hardware to understand exactly what your software is managing. The team is small enough that your design decisions shape the architecture, and large enough that you always have experts to learn from across software, hardware, and operations.

The Ideal Candidate

You are a strong software engineer first: you design clean abstractions, write high-quality, testable code, and reason about correctness, scalability, and failure modes. You care about the customer of your software, whether that is an operations engineer, a partner service, or an internal tool user. You communicate clearly, raise the engineering bar for your team, and can partner with hardware, firmware, and operations engineers to build software that reflects how the fleet actually behaves.

Key job responsibilities

Service & Platform Development

1. Design, build, and operate scalable distributed services and control planes that automate fleet provisioning, health evaluation, and remediation across a large scale of servers

2. Design APIs, data models, and system architectures that balance latency, availability, durability, and cost at fleet scale

3. Own software end to end: design, implementation, testing, deployment, and production operations, following engineering best practices for code review, source control, and CI/CD

Data & Telemetry Platforms

1. Build data and telemetry pipelines that ingest, correlate, and transform signals from across the fleet into actionable insights and automated actions

2. Develop analytics and reporting services, dashboards, and alerting that give engineering and operations teams real-time visibility into fleet health

3. Apply appropriate algorithms and data structures to process high-volume telemetry efficiently and reliably

Reliability & Operational Excellence

1. Instrument services for observability; participate in post-incident reviews, identify contributing causes, and drive permanent software fixes that eliminate whole classes of risk

2. Improve the resilience, performance, and cost of the services you own, and reduce operational load through automation

3. Produce clear, maintainable documentation for services, APIs, and runbooks so others can operate and extend your work

Technical Leadership & Collaboration

1. Decompose complex problems into well-scoped components that you and other engineers can deliver in parallel

2. Partner with systems, firmware, hardware, and operations engineers to translate fleet behavior into software requirements and correct abstractions

3. Review designs and code, mentor engineers, and set engineering standards for the team

May require occasional (<10%) regional and international travel to Design and Manufacturing Partner sites.

A day in the life

You start the day reviewing a design document for a new remediation control-plane API, leaving feedback on the data model and failure handling. Mid-morning, you pair with a teammate to debug elevated latency in the telemetry ingestion service and ship a fix. In the afternoon, you review overnight metrics from a service you own, then write code for a new fleet-health scoring pipeline. You end the day in a design review with a partner team, aligning on the API contract your service will expose to their tooling.

About the team

The Hardware Engineering AI/ML UltraServer platform team is a group of engineers and technical program managers directly responsible for launching GPU-accelerated servers into the AWS fleet. Located in Seattle, Austin, and Cupertino, we collaborate with global development teams and ODM partners to deliver next-generation AI/ML infrastructure deployed in datacenters worldwide. We move fast with small, empowered teams delivering end-to-end, from server conception through fleet-scale operations.