Description

Want to help drive the success of Machine Learning technologies at AWS? We seek a Software Development Engineer II for the ML Infrastructure team to build the platforms that guarantee top performance of AWS ML and HPC technologies. Our performance data directly influences launch
decisions for new EC2 instance types and has visibility at senior leadership.

Join us as we expand the AWS offerings for AI, including Trainium, Neuron and the Elastic Fabric Adapter (EFA). You'll build CI/CD systems, orchestrate GPU clusters, create performance dashboards, and develop AI-powered automation - all to ensure latest ML networking software ships with confidence.

Key job responsibilities
Build and maintain infrastructure that monitors and reports on functionality and performance of massive testing workloads run at scale across multiple GPU instance types. Use Jenkins, internal Amazon CI/CD tools, Linux, and public AWS products to automate testing and delivery of ML networking libraries - including collective communication frameworks, network transport layers, and GPU communication libraries. Write Python code that orchestrates large clusters, runs benchmarks and ML applications across a matrix of instance types, operating systems, and
software stack versions. Use AWS Managed Grafana and Athena to digest performance data and build dashboards that catch functional and performance regressions before they reach customers. Build automation using LLMs to analyze test failures and surface actionable insights to developers. Contribute to cross-team readiness for new instance type launches by delivering performance data that shapes go/no-go decisions. Manage the complexity of infrastructure covering many instance types, software stacks, Linux operating systems, and latest releases
and make it easy to evolve.

A day in the life
You write Python to orchestrate test workloads across large GPU clusters and TypeScript with CDK to ensure all infrastructure is code, reviewed and committed to automated pipelines. You manage shared development clusters using SLURM and AWS ParallelCluster, supporting multiple teams while keeping costs down. You build automation that analyzes nightly test results and surfaces regressions to developers. You write crisp designs for your projects, communicating clearly to your peers what you will build and why.

About the team
We are part of Annapurna Labs, a subsidiary in AWS that builds software and hardware that make ML on EC2 work. Our organization is a dedicated group of innovators that have invented new networks, new silicon, new software suites, and combined those to entice customers to move immense ML and HPC workloads to the cloud. The ML Infrastructure team is laser focused on making AWS the best and most cost-effective place for customers to do AI at scale. Our work directly enables the launch of new GPU instance types.