Lead GPU Performance & HPC Systems Engineer

Nebius Group; United States; Remote; USD 170,000 - 300,000; Full time

Benefits offered by this job

Health insurance

401(k) plan

Parental leave

Remote work reimbursement

Disability & life insurance

Job summary

Nebius is hiring a Lead Software Systems Engineer - GPU Performance to optimize large-scale GPU clusters spanning hardware and software. You will analyze system behavior, identify bottlenecks, and implement tunings that improve performance for training and inference workloads.

The role covers hardware interactions, kernel-level work, and collaboration with vendors and internal teams to meet aggressive performance targets in a global AI cloud platform.

Qualifications

• 5+ years of system-level software development focused on performance optimization.
• 3+ years hands-on Linux systems administration, troubleshooting and tuning.
• Deep understanding of server architecture including PCIe, NICs, kernel, and HPC.
• Proficient in C/C++, Go, or Python for performance-oriented tasks.

Responsibilities

• Identify bottlenecks across hardware and software stack and drive improvements.
• Troubleshoot GPU cluster performance under real workloads (training and inference).
• Evaluate new hardware, configurations, and tuning approaches via the software stack.
• Support escalations from internal teams and customers on performance issues.
• Collaborate with infra, software, and hardware vendor teams (NVIDIA, Mellanox, Intel).
• Contribute to hardware and cluster qualification to meet performance targets.

Skills

System-level software development

Linux systems

PCIe architecture

Networking (InfiniBand/RoCE)

High-performance computing

Performance-oriented languages

Job description

Nebius is hiring a Lead Software Systems Engineer - GPU Performance to optimize large-scale GPU clusters spanning hardware and software. You will analyze system behavior, identify bottlenecks, and implement tunings that improve performance for training and inference workloads.

The role covers hardware interactions, kernel-level work, and collaboration with vendors and internal teams to meet aggressive performance targets in a global AI cloud platform.