Amazon
Software Development Engineer II, AWS SageMaker AI
$144K–$194K
Bellevue, WA · Full-time · Posted Aug 5, 2026
- Experience
- 3+ years non-internship software development; 2+ years non-internship design or architecture; 1+ year software development engineer or related occupational experience; 1+ year large-scale or distributed software development using a listed language; 1+ year Object Oriented Design.
- Education
- Bachelor's degree or foreign equivalent in Computer Science, Engineering, Mathematics, or a related field.
About the job
Build and operate Model Factory distributed systems that orchestrate foundation-model training and customization workflows on AWS.
What you'll do
- Design, build, test, deploy, monitor, and operate workflow orchestration services and platform components end-to-end.
- Integrate compute backends and improve reliability, recovery, and observability for large-scale training jobs.
- Contribute to the SDK, schema and contract layer, CDK infrastructure, technical designs, and collaboration with ML teams.
What they ask for
- 3+ years of non-internship professional software development experience.
- 2+ years of non-internship system design or architecture experience.
- At least 1 year designing and developing large-scale or distributed applications using C#, C++, Java, or Perl; and at least 1 year of Object Oriented Design experience.
- Bachelor's degree or foreign equivalent in Computer Science, Engineering, Mathematics, or a related field; experience programming with at least one software language.
Nice to have
- 3+ years of full software development life cycle experience.
- Experience with orchestration, containers and cluster compute, cloud infrastructure-as-code, managed cloud services, or distributed ML training.
Pay
143,700.00 - 194,400.00 USD annually (base salary)
Tools
SageMaker, EMR, AWS Batch, Kubernetes/EKS, PyTorch/FSDP, verl, NeMo/Megatron, AWS CDK
AI focus
Build the platform for preparing, training, customizing, evaluating, and deploying foundation models, including integrations for distributed training and accelerator compute.
Model-written summary · the original posting is authoritative