Open Source ML Systems Engineer for AMD GPUs

Advanced Micro Devices, Inc.; San Jose (CA); On-site; USD 160,000 - 240,000; Full time

Benefits offered by this job

AMD benefits

Collaborative open source environment

Job summary

Advanced Micro Devices, Inc. in San Jose, CA is seeking an Open Source Software Engineer to build and optimize ML systems that run on AMD hardware. You will bridge open source AI software with GPU systems, design, implement, test, and upstream contributions, working close to hardware while integrating with ML frameworks.

The role emphasizes performance, kernel fusion, and collaboration in open source projects, with rigorous testing and validation to ensure reliability across AMD platforms.

Qualifications

• Strong familiarity with LLM systems, inference workloads, compute and optimization.
• Experience optimizing GPU kernels, particularly for AMD architectures.
• Ability to learn multiple kernel programming languages and models.
• Understanding of kernel fusion, GPU execution, and pipelined execution.

Responsibilities

• Develop and maintain open source integrations between ML/AI software and AMD hardware.
• Optimize kernels and critical execution paths for AMD GPUs.
• Profile workloads, identify bottlenecks, and implement measurable optimizations.
• Contribute high-quality code upstream to open source projects.
• Write comprehensive unit, integration, and performance tests.

Skills

LLM systems

GPU kernel optimization

Open source

CI and testing

Kernel fusion

Education

Bachelor's/Master's in CS/CE/EE

Tools

ROCm

CUDA

SGLang

vLLM

Job description

Advanced Micro Devices, Inc. in San Jose, CA is seeking an Open Source Software Engineer to build and optimize ML systems that run on AMD hardware. You will bridge open source AI software with GPU systems, design, implement, test, and upstream contributions, working close to hardware while integrating with ML frameworks.

The role emphasizes performance, kernel fusion, and collaboration in open source projects, with rigorous testing and validation to ensure reliability across AMD platforms.