Senior AI Triton Distributed GPU Engineer

AMD; United States; On-site; USD 150,000 - 210,000; Full time

Job summary

AMD is seeking a seasoned engineer to advance Triton distributed execution and communication on AMD Instinct GPUs. You will work across compiler, runtime, and hardware layers to enable scalable multi-GPU training and inference, with close collaboration across architecture, software, and performance teams.

You will design native distributed capabilities, optimize inter-GPU data movement, and contribute to open-source Triton and ROCm ecosystems.

Qualifications

• Experience in compiler development and GPU software.
• Proven able to optimize workloads across multi-GPU systems.
• Strong understanding of GPU memory hierarchy and interconnects.
• Familiarity with distributed GPU communication libraries.

Responsibilities

• Design and develop native distributed communication and execution within Triton AMDGPU backend.
• Implement compiler/runtime mechanisms for GPU-initiated communication and collective ops.
• Optimize inter-GPU data movement and scheduling for AI workloads.
• Collaborate with architecture, compiler, runtime and performance teams.

Skills

Compiler development

GPU software

Distributed systems

Performance engineering

Triton

Runtime

GPU architectures

Inter-GPU interconnects

Tools

Triton compiler

ROCm

NCCL

Job description

AMD is seeking a seasoned engineer to advance Triton distributed execution and communication on AMD Instinct GPUs. You will work across compiler, runtime, and hardware layers to enable scalable multi-GPU training and inference, with close collaboration across architecture, software, and performance teams.

You will design native distributed capabilities, optimize inter-GPU data movement, and contribute to open-source Triton and ROCm ecosystems.