Location
Remote position based in the United States; candidates must be authorized to work in the country where they apply.
About the company
A global AI cloud platform provider building full-stack infrastructure for developers and enterprises, from data and model training to production deployment. The company operates large-scale GPU clusters and works across compute, storage, networking, and applied AI, with a distributed team of over 1,500 people.
Responsibilities
- Analyze and optimize performance of large-scale GPU clusters across hardware, system software, networking, virtualization, and distributed communication layers.
- Diagnose bottlenecks under real training and inference workloads, and drive tuning and validation improvements for how clusters are built and operated.
- Evaluate and integrate new hardware, system configurations, and tuning approaches through the software stack.
- Support complex performance escalations and collaborate with infrastructure, software, and hardware vendor teams on cluster qualification.
Requirements
- 5+ years of system-level software development focused on performance optimization and low-level programming.
- 3+ years of hands-on Linux administration, troubleshooting, and performance tuning.
- Deep knowledge of server architecture, including PCIe devices, NICs, Linux OS/kernel, and HPC systems.
- Strong proficiency in C/C++, Go, or Python, with coding interviews as part of the process.