Software Engineer

Super Micro Computer, Inc. • San Jose, CA

37d • $153,000 - $167,000 • Onsite

About The Position

This role contributes directly to the development of GPU-accelerated software used in internal tooling pipelines and in customer-facing solution deployments. The engineer produces optimized code paths targeting NVIDIA and AMD GPUs, implements kernel-level improvements, and integrates software components with hardware engineering platforms. The position operates in a multi-vendor environment, supporting performance-critical workloads that require precise tuning, profiling, and validation across architectures. This role will be based in our headquarters located in San Jose, CA.

Requirements

• BS in EE/CS/CIS
• 3-5 years of experience.
• Proficiency with CUDA, HIP, GPU compute fundamentals, and parallel programming constructs.
• Experience with PyTorch, TensorFlow, or comparable ML frameworks for GPU-backed execution.
• Competence with kernel profilers and performance analysis toolchains (Nsight Systems/Compute, rocprof, OmniPerf).
• Understanding of GPU memory models, shared-memory usage, warp/wavefront execution, and optimization for latency and throughput.
• Strong C/C++ and Python development capability with clean, maintainable coding practices.
• Familiarity with Linux development environments, build systems, and driver-level debugging fundamentals.

Nice To Haves

• Master degree in EE/CS/CIS
• Exposure to ROCm, Triton, OpenCL, or domain-specific kernel DSLs.
• Understanding of GPU virtualization, containerized execution environments, and distributed or multi-GPU communication (NVLink, Infinity Fabric, PCIe topology).
• Experience in customer-oriented engineering roles, pre/post-sales technical support, or solution-integration environments.
• Knowledge of mixed-precision strategies and architecture-specific tuning for inference or training workloads.
• Background working with system-level performance bottlenecks involving memory bandwidth, NUMA layouts, and accelerator interconnects.

Responsibilities

• Implement and optimize GPU-accelerated code using CUDA, HIP, and vendor SDKs.
• Port workloads and kernels between NVIDIA and AMD GPU platforms with minimal regression.
• Develop internal benchmarking suites, diagnostics, and performance tooling for engineering teams.
• Collaborate with hardware engineering, solution architects, and customer-facing groups to align software behavior with system-level constraints.
• Profile, debug, and validate performance of GPU workloads using Nsight, rocprof, OmniPerf, and related tools.
• Maintain clear documentation for kernels, toolchains, and multi-GPU execution paths.
• Contribute to continuous integration pipelines for GPU-targeted builds and tests.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Industry: Computer and Electronic Product Manufacturing
• Number of Employees: 5,001-10,000 employees