Senior AI/ML Systems Engineer for Accelerator Optimization

Socket.dev; Seattle (WA); On-site; USD 168,000 - 227,000; Full time

Job summary

Amazon, the maker of AWS Neuron, seeks a Senior Software Engineer to optimize AI model inference on custom accelerators like Inferentia and Trainium. You will work across compiler, runtime, and hardware teams to push the performance of Generative AI workloads and ML systems at scale.

You will design high-performance kernels, profiling and tuning ML workloads, and collaborating with customers and open-source communities to deliver optimized solutions on AWS accelerators.

Qualifications

• Bachelor's degree in CS or related field required.
• 5+ years of professional software development experience.
• Proficiency in Python and/or C++.
• 5+ years of design/architecture leadership for systems.
• Experience debugging, profiling, and optimizing large-scale systems.
• Strong knowledge of memory management and parallel computing.
• Experience owning a performance optimization roadmap.

Responsibilities

• Design, develop, and optimize ML models on custom AI accelerators.
• Lead in distributed inference support for PyTorch in Neuron SDK.
• Profile and optimize performance across Neuron hardware generations.
• Build infrastructure to onboard multiple models with diverse architectures.
• Develop high-performance kernels for ML operations on Neuron hardware.
• Collaborate across teams to enable and optimize customer ML workloads.

Skills

Python programming

C++ programming

System performance

Debugging and profiling

Parallel computing

Education

Bachelor's degree

Master’s degree

Tools

PyTorch

CUDA

Triton

NCCL

Job description

Amazon, the maker of AWS Neuron, seeks a Senior Software Engineer to optimize AI model inference on custom accelerators like Inferentia and Trainium. You will work across compiler, runtime, and hardware teams to push the performance of Generative AI workloads and ML systems at scale.

You will design high-performance kernels, profiling and tuning ML workloads, and collaborating with customers and open-source communities to deliver optimized solutions on AWS accelerators.