AI Compute Performance Engineer: GPU & HPC

Cisco Systems, Inc; Milpitas (CA); On-site; USD 155,000 - 223,000; Full time

Benefits offered by this job

Medical, dental, vision insurance

401(k) with company match

Paid parental leave

Disability coverage

Life insurance

Restricted Stock Units

Job summary

Cisco Systems, Inc in Milpitas, CA is seeking a Software Engineer to analyze and improve the performance of AI compute systems across hardware, system software, and application workloads.

You will run AI training and inference workloads on GPU-accelerated systems, investigate bottlenecks, develop automation for performance testing, and collaborate with hardware and software teams to deliver validated performance results.

Qualifications

• Debugging, testing, or tuning Linux-based systems.
• Experience running or evaluating AI, ML, or HPC workloads.
• 3+ years of programming in Python, C, C++, Go, or Bash.

Responsibilities

• Run and analyze AI training, inference, and HPC workloads on GPU-accelerated systems.
• Investigate compute, memory, PCIe, storage, and workload scaling issues.
• Use telemetry, logs, traces, and profiling tools to identify performance bottlenecks.
• Develop automation for performance testing, regression detection, and failure triage.
• Work with hardware and software teams to resolve system and workload performance issues.
• Document findings, test procedures, and performance results.

Skills

Python

C++

Go

Bash

Linux

CUDA

Education

Bachelor’s degree with 5+ years of related experience

Master’s degree with 3+ years of related experience

PhD with relevant research experience

Tools

Nsight

DCGM

rocprofiler

Kubernetes

Slurm

NVLink/NVSwitch

Job description

Cisco Systems, Inc in Milpitas, CA is seeking a Software Engineer to analyze and improve the performance of AI compute systems across hardware, system software, and application workloads.

You will run AI training and inference workloads on GPU-accelerated systems, investigate bottlenecks, develop automation for performance testing, and collaborate with hardware and software teams to deliver validated performance results.