Description

The Software Development Engineer II will design, build, and maintain cloud-based repair and recovery workflows for NVIDIA GB200 / GB300 UltraServers, orchestrating repair and recovery operations from impairment detection through completed recovery. This role requires expertise in AWS services, system architecture, and cross-functional collaboration with Capacity Management, Hardware Engineering, and Datacenter Operations to manage AI/ML infrastructure.

Key job responsibilities
The Software Development Engineer (SDE II) on the EC2 UltraServer Availability team is responsible for ensuring high availability of customer GB200 and GB300 UltraServers by orchestrating complex repair and recovery workflows. Following are the core responsibilities

System Design & Architecture

* Design and architect solutions that are cross-functional to Capacity Management, Hardware Engineering, and Datacenter Operations
* Work in environments where the technology strategy is defined but the solution design is not
* Build solutions that are stable, logical, testable, and efficient with the ability to independently make trade-off decisions
* Investigate and develop design concepts to frame solution sets at an application and product level

Software Development

* Build cloud-based solutions using AWS native services for scaling infrastructure frameworks
* Write high-quality, maintainable code with proper testing and code reviews
* Develop and maintain the repair and recovery workflows for GB200 and GB300 UltraServer hosts
* Implement automation for diagnostic triage, hardware testing, cable validation, and testing processes
* Create observable systems with appropriate metrics and alarming

Operational Excellence

* Execute and monitor UltraServer workflows for UltraServer repair
* Troubleshoot workflow failures and coordinate with downstream teams
* Focus on operational excellence by identifying problems and proposing solutions

Hardware & Software Integration

* Work with hardware and software integrations specific to GPU clusters and AI/ML training systems
* Manage network partition configurations for multi-node GPU clusters
* Handle firmware validation and consistency checks across asset groups

Team Collaboration

* Collaborate with customers and stakeholders to convert business needs into technical designs
* Participate in code reviews and technical assessments


A day in the life
This is a hands-on position in which you will own everything from end to end: requirements gathering, designs, design reviews, implementations, code reviews, incremental feature launches, operations, mentoring, and the driving of continuous improvement.

About the team
The EC2 UltraServer Availability team is a high-performing engineering organization responsible for maintaining high availability of NVIDIA-based ML infrastructure at scale. We manage end-to-end repair and recovery workflows for GB200 and GB300 UltraServers, from initial problem detection through repair and recovery. Our team drives operational excellence through continuous improvement of problem detection, repair efficacy, and customer impact mitigation. We work closely with hardware engineering, data center operations, and EC2 service teams to ensure reliable, efficient recovery of critical ML compute capacity. This is a high-impact role leading a two-pizza team of talented engineers solving complex technical challenges in one of Amazon's fastest-growing infrastructure domains.