PhD AI Systems & GPU Performance Intern - Co-op (Hybrid)

AMD; Santa Clara (CA); Hybrid; USD 55,000 - 96,000; Full time

Job summary

AMD is seeking a PhD AI Systems & GPU Performance Engineering Intern to optimize state-of-the-art AI models and workloads on AMD Instinct GPUs. You will work with ROCm, PyTorch, JAX and related tools to profile, benchmark, and improve performance.

You will learn to develop reproducible benchmarking frameworks, analyze memory bandwidth, kernel execution, and scheduling strategies, and collaborate with researchers to drive efficiency and scalability across AI workloads.

Qualifications

• Pursuing a PhD with expected graduation after the internship.
• Experience with PyTorch/JAX/TensorFlow and large-scale AI models.

Responsibilities

• Profile, benchmark, and optimize AI training and inference workloads on AMD Instinct GPUs using ROCm ecosystem.
• Develop reproducible benchmarking frameworks and automation scripts for performance validation across models and stacks.
• Analyze and optimize end-to-end AI workflows including LLMs and generative AI, using model optimization techniques.
• Collaborate with engineers to evaluate GPU performance and provide data-driven optimization recommendations.
• Explore cutting-edge AI systems research on AMD hardware to advance training and inference tech.

Skills

Python

C/C++

GPU programming

Linux

Profiling

Education

PhD candidate in CS/CE/EE

Tools

rocProfiler

ROCm

Nsight

Omniperf

Job description

AMD is seeking a PhD AI Systems & GPU Performance Engineering Intern to optimize state-of-the-art AI models and workloads on AMD Instinct GPUs. You will work with ROCm, PyTorch, JAX and related tools to profile, benchmark, and improve performance.

You will learn to develop reproducible benchmarking frameworks, analyze memory bandwidth, kernel execution, and scheduling strategies, and collaborate with researchers to drive efficiency and scalability across AI workloads.