AI/ML Software Engineer for High-Performance Training

Amazon; Cupertino (CA); On-site; USD 165,000 - 224,000; Full time

Job summary

Amazon Neuron Training team is hiring a Software Engineer- AI/ML to optimize distributed training on Trainium. You will work across the Neuron stack, from PyTorch frameworks to the compiler and runtime, to push performance and efficiency for frontier-scale models.

You will shape parallelism strategies, profile workloads, and collaborate with compiler, runtime, and framework teams to land impactful improvements across large-scale training environments.

Qualifications

• 3+ years of non-internship professional software development experience.
• 2+ years of non-internship design or architecture experience.
• Experience programming with at least one software programming language.

Responsibilities

• Lead efforts to optimize distributed training performance on Trainium across the Neuron software stack.
• Collaborate across PyTorch and Neuron stack to enable and tune large-scale training workloads on Trainium instances.
• Own parallelism strategies spanning data, tensor, pipeline, expert, and context parallelism; apply reduced-precision formats where beneficial.
• Profile end-to-end performance to identify bottlenecks and drive fixes with compiler, runtime, and collectives teams.
• Translate performance gaps into framework requirements and contribute upstream to open source frameworks.

Skills

Software development

System design/architecture

Programming languages

Education

Bachelor's degree in Computer Science or equivalent

Tools

PyTorch

JAX

Trainium

Job description

Amazon Neuron Training team is hiring a Software Engineer- AI/ML to optimize distributed training on Trainium. You will work across the Neuron stack, from PyTorch frameworks to the compiler and runtime, to push performance and efficiency for frontier-scale models.

You will shape parallelism strategies, profile workloads, and collaborate with compiler, runtime, and framework teams to land impactful improvements across large-scale training environments.