Advanced AI Workloads Performance and Scalability Engineer

Quick Facts

• Role: Performance and Scalability Engineer for advanced AI workloads

• Focus: AI inference performance optimization on AMD GPUs (kernel + systems end-to-end)

Description

You will drive performance optimization for leading AI models on AMD GPUs, working end-to-end across the stack—from profiling and diagnosing bottlenecks to optimizing customer-relevant serving configurations (e.g., agentic coding, long-context, high-throughput serving). The work includes kernel and systems-level improvements, custom kernel integration into serving frameworks, and multi-node distributed inference optimizations. You will lead technically, measure everything, and present measurable results to customer engineering leadership.

Responsibilities

• Drive end-to-end performance optimization across kernel and systems layers to close competitive gaps

• Profile, diagnose, and resolve cross-stack bottlenecks from GPU kernels and operator dispatch to framework scheduling and multi-node communication

• Use profiling tools to identify issues such as occupancy limits, L2 cache thrashing, register pressure, and memory coalescing problems

• Lead customer-facing technical engagements and deliver measurable performance uplifts

• Integrate and optimize custom kernels (Triton, Gluon, CK, PyDSL, ASM, AITER) within serving frameworks, including dispatch paths, shape extraction, and backend selection

• Optimize multi-node distributed inference, including communication-compute overlap and scale-out performance

• Develop and refine reusable performance optimization methodology across the team

• Leverage AI agents to accelerate daily performance engineering workflows

• Upstream optimizations into open-source frameworks such as vLLM, SGLang, and PyTorch

Requirements

• Software development experience in GPU computing, AI systems, or high-performance computing

• Deep hands-on experience with AI serving frameworks and their internals (e.g., vLLM, SGLang, TensorRT-LLM or similar)

• Strong end-to-end profiling and bottleneck diagnosis ability from user request to GPU kernel

• Deep understanding of GPU kernel performance characteristics (occupancy, register/LDS pressure, memory coalescing, cache utilization, wavefront scheduling, instruction-level bottlenecks)

• Ability to read kernel profiling data and translate it into concrete optimization actions

• Understanding of model architectures and inference paradigms and how they map to hardware

• Experience integrating or developing custom kernels (HIP, CUDA, Triton, CK, or similar)

• Understanding of multi-GPU and multi-node distributed systems (scale-up/scale-out, RCCL/NCCL, RDMA, communication-compute overlap)

• Strong proficiency in Python and C++

• Strong Linux systems knowledge

• Excellent written and verbal English communication skills

• Customer-facing technical leadership experience

Benefits

Benefits are described as “AMD benefits at a glance.”