Senior GPU Inference Performance Engineer

Amd

Confirmed live today High trust

Hybrid

Quick summary

Work type: Hybrid
Location: Santa Clara, CA
Salary: $164,000–$246,000 / yr
Posted: 109 days ago
Freshness: Confirmed live today
Closes: Jul 9, 2027
Nearby: 99+ roles within 25 mi

Market check

Salary context

Competitive pay

How this pay compares to similar roles

Similar $211k

This role $205k

$144k $271k

below market most similar roles pay here above market

This role pays more than 50% of similar roles. Most pay $186,725–$235,750 — the blue band above. At the midpoint, this role pays about $205k versus about $211k for comparable roles.

Based on 240 similar postings.

Employer

About Amd

AMD (Advanced Micro Devices) is a semiconductor company that develops high-performance processors, graphics cards, and adaptive computing solutions for gaming, data centers, and embedded markets. Industry: Semiconductors

Amd currently has 496 open roles on FindRole.

Listed pay typically runs $172,000–$258,000 across 496 roles with salary data.

Most-posted roles

View all roles at Amd

At a glance

TL;DR · Senior GPU Inference Performance Engineer

Role Posting Log in to save

The Senior GPU Inference Performance Engineer joins the team to own end-to-end performance analysis of GPU-accelerated AI inference workloads. The role involves profiling, diagnosing, and explaining performance across the full stack, including GPU silicon, communication libraries, networking fabrics, operating systems, and software runtimes. Key responsibilities include full-stack GPU profiling, systems and runtime performance analysis, and conducting competitive performance analysis against other accelerator vendors. The engineer will optimize multi-server inference networking and the Kubernetes stack while building benchmarking harnesses and automation tools. Required technical skills include proficiency with ROCm, CUDA, NCCL/RCCL, RDMA/RoCE, and Linux systems. The position addresses the technical challenge of identifying bottlenecks in HBM bandwidth, compute utilization, and kernel scheduling to improve throughput and latency for large language model workloads.

Skills

ROCm CUDA Python C++ Kubernetes RDMA RoCE NCCL RCCL Docker Linux HIP MPI UCX Nsight Systems rocProfiler Omniperf HPC Distributed Systems

What you'll do

Profile and analyze GPU inference workloads to identify bottlenecks in HBM bandwidth, compute utilization, and memory allocation.
Diagnose performance interactions between GPU runtimes, Linux operating systems, device drivers, and container runtimes.
Execute head-to-head benchmarks against competitors to explain performance gaps using hardware and software evidence.
Optimize multi-server inference topologies by profiling RDMA/RoCE traces and collective communication libraries like NCCL/RCCL.
Analyze and resolve performance overheads and jitter within the Kubernetes GPU operator and container runtime stacks.
Build reproducible benchmarking harnesses, profiling scripts, and automated performance regression dashboards.
Present data-backed performance findings and technical reports to product and executive stakeholders.

What we're looking for

Perform end-to-end performance analysis of GPU-accelerated AI inference workloads across the full stack.
Profile and diagnose performance across GPU silicon, communication libraries, networking fabrics, operating systems, and software runtimes.
Design and execute head-to-head benchmarks against competitors to explain performance differences using hardware and software evidence.
Profile and optimize multi-server inference topologies including RDMA/RoCE networking and NCCL/RCCL collective profiling.
Analyze performance overheads of GPU operators, container runtimes, and Kubernetes scheduling on inference latency.
Build reproducible benchmarking harnesses, profiling scripts, and performance regression dashboards.
Background in GPU performance engineering, HPC, distributed systems, networking, or operating systems (preferred).
Bachelor's degree in Computer Science, Computer Engineering, Electrical Engineering, or a related technical field (preferred).

Save