Senior Staff GPU Datacenter Performance Engineer
- Location
- Santa Clara, CA
- Employment
- Full Time
- Work model
- On-Site
- Level
- Staff
- Posted
Skills
About this role
ADVANCE YOUR CAREER. ADVANCE THE WORLD. At AMD, we believe technology has the power to solve the world’s most important challenges. From advancing healthcare and scientific discovery to powering AI and the technologies people rely on every day, innovation at AMD is shaping the future. Whether you’re designing next-gen processors, enabling AI breakthroughs, or bringing leading edge products to market, every role at AMD contributes to something bigger — technology that moves the world forward. Join us and, together, we’ll advance your career.
THE ROLE
The successful candidate will assume responsibility for post-silicon activities related to performance characterization and optimization of AMD Datacenter products, spanning both single-node and multi-node deployments. THE PERSON: You have a passion for computer architecture, along with strong knowledge of HPC and ML workloads. You are a team player with excellent communication skills and experience collaborating with engineers across multiple sites and time zones. You are a quick learner, self-starter, and able to drive multiple complex tasks to completion.
KEY RESPONSIBILITIES
Define and lead performance-characterization strategies for AMD DataCenter products, spanning HPC and ML workloads and single-node through multi-node deployments. Establish scalable, repeatable workload-execution and performance-data-collection frameworks. Influence product power, thermal, and performance definition by connecting architectural goals to measured silicon and system behavior. Lead the development and evolution of characterization and optimization test plans across diverse usage scenarios. Drive performance attainment across scale-up and scale-out configurations, including: Multi-GPU scaling efficiency within a node Interconnect bandwidth utilization, including XGMI / Infinity Fabric Collective communication efficiency and communication–compute overlap Rack- and cluster-level bottleneck identification and mitigation Analyze interactions among power-management features, thermal behavior, and performance; develop configurations that balance throughput and performance-per-watt. Identify architectural and system-level bottlenecks, and design experiments that expose and help mitigate worst-case performance scenarios. Lead prototyping and experimentation to evaluate architectural, firmware, driver, and platform enhancements. Drive complex root-cause investigations across hardware, firmware, drivers, platform, and workload stacks through to resolution and performance closure. Serve as a technical authority across teams; communicate findings, tradeoffs, and recommendations to engineering stakeholders. Mentor engineers and raise the technical quality and consistency of post-silicon performance methodologies, tools, and workflows.
PREFERRED EXPERIENCE
Demonstrated technical leadership on complex, cross-functional engineering programs, including influence across organizations, locations, and time zones. Strong programming skills, preferably in Python; experience with ML frameworks such as TensorFlow or PyTorch is an advantage. Proficiency in C/C++ and shell scripting, with experience in performance tooling and automation. Deep understanding of computer architecture, GPU system organization, and performance analysis. Extensive knowledge of HPC and ML workloads, including scaling behavior, collective communications, and performance bottlenecks. Experience analyzing scale-up and scale-out performance at rack or cluster level. Familiarity with high-speed I/O (HSIO) performance, such as PCIe, is a plus. Strong analytical judgment and problem-solving skills, with a record of turning ambiguous performance symptoms into evidence-based conclusions and actionable fixes. Excellent written and verbal communication skills; able to explain complex technical findings and influence engineering decisions. ACADEMIC CREDENTIALS: Bachelor's or Master's