Advanced AI Workloads - Performance and Scalability Engineer

Advanced Micro Devices, Inc•San Jose, CA

About The Position

AMD is seeking a performance-focused engineer to drive AI inference performance on AMD GPUs. The role leads a small, highly technical team and works end-to-end across the stack to profile, diagnose, and optimize leading models on customer-relevant serving configurations, including agentic coding, long-context, and high-throughput serving. The engineer tackles performance challenges across strategic customer engagements and delivers measurable improvements and reusable methodologies. The role involves GPU kernel and systems-level optimization, distributed inference, customer-facing technical leadership, integration of custom kernels into serving frameworks, and upstream contributions to open-source projects. The engineer is expected to lead through technical depth, use AI agents and tools in daily work, and establish effective AI-assisted performance-engineering practices. The preferred location is San Jose, California. This position is not eligible for visa sponsorship.

Requirements

Software development experience in GPU computing, AI systems, or high-performance computing.
Deep hands-on experience with AI serving frameworks such as vLLM, SGLang, TensorRT-LLM, or similar, including their internals.
Strong background in end-to-end workload profiling and bottleneck diagnosis, with the ability to trace a request through to GPU kernels and back.
Understanding of GPU kernel performance characteristics, including occupancy, register and LDS pressure, memory coalescing, cache utilization, wavefront scheduling, and instruction-level bottlenecks.
Ability to read and reason about kernel-level profiling data and translate it into concrete optimization actions.
Understanding of model architectures such as transformers, MoE, and diffusion; inference paradigms such as speculative decoding, prefill-decode disaggregation, and continuous batching; and how they map to hardware.
Experience with custom kernel development or integration using HIP, CUDA, Triton, CK, or similar.
Understanding of multi-GPU and multi-node distributed systems, including scale-up and scale-out topologies, RCCL/NCCL, RDMA, and communication-compute overlap.
System and rack-level design awareness, including performance tradeoffs across the deployment stack.
Strong proficiency in Python and C++.
Customer-facing technical leadership experience, including engaging customers, presenting findings, and driving decisions.
Daily use of AI agents and tools, with a mindset toward defining AI-powered workflows.
Strong Linux systems knowledge.
Excellent written and verbal English communication skills.
Bachelor's, Master's, or PhD in Computer Science, Computer Engineering, Electrical Engineering, or equivalent; equivalent exceptional industry experience is valued.

Nice To Haves

Advanced degree in Computer Science, Computer Engineering, Electrical Engineering, or an equivalent field.
Experience with serving frameworks such as vLLM, SGLang, or TensorRT-LLM.
Experience with custom kernel technologies such as HIP, CUDA, Triton, or CK.

Responsibilities

Drive end-to-end performance optimization across the stack for leading models and customer-relevant serving configurations, closing competitive gaps through kernel and systems-level optimizations.
Profile, diagnose, and resolve cross-stack performance bottlenecks involving GPU kernels, operator dispatch, framework-level scheduling, and multi-node communication.
Use profiling tools to diagnose kernel performance issues, including occupancy limitations, L2 cache thrashing, register pressure, and memory coalescing issues, and translate findings into actionable optimizations.
Lead customer-facing technical engagements by presenting findings, recommending optimizations, and delivering measurable performance improvements.
Integrate and optimize custom kernels, including Triton, Gluon, CK, PyDSL, ASM, and AITER, within serving frameworks; understand dispatch paths, shape extraction, and backend selection.
Optimize multi-node distributed inference, including communication-compute overlap, parallelism strategies, and scale-out performance.
Develop and refine shared performance-optimization methodology across the team.
Use AI agents to accelerate daily work and define best practices for AI-assisted performance engineering.
Upstream optimizations into open-source frameworks such as vLLM, SGLang, and PyTorch.