Senior GPU Compute Architect

OneClick Smart Resume; San Mateo (CA); On-site; USD 345,000 - 399,000; Full time

Job summary

Roblox is seeking a Principal Software Engineer, GPU Compute, in San Mateo, CA to lead GPU compute initiatives and own the machine management layer for production-scale accelerators. You will partner with Kubernetes, networking, and cloud teams to drive end-to-end GPU strategy and reliability.

The role requires deep GPU expertise, strong Go skills, and experience operating GPU/AI workloads in production. Onsite work in our San Mateo office is expected with equity and benefits available.

Qualifications

• 10+ years of experience building and operating large-scale distributed systems and infrastructure.
• Deep, hands-on GPU expertise at the machine management layer and above: GPU host provisioning, driver and firmware lifecycle, GPU health and reliability.
• Proven track record as compute expert with scalable GPU infrastructure.
• Strong proficiency in Go or other well-structured programming languages.
• Experience operating GPU and AI workloads in production, including CUDA, GPU scheduling, and high-performance networking (NVLink, InfiniBand, RoCE).
• Familiarity with Kubernetes for GPU workloads and bare-metal concepts (firmware, BMC/IPMI/Redfish, OS imaging).
• Anchor expert with leadership to uplift engineers around you.

Responsibilities

• Serve as the GPU technical leader for the Compute team across Kubernetes, Machine Bootstrap, Networking, and Cloud.
• Own GPU host lifecycle: driver, firmware, CUDA stack management, health, and remediation of GPU-specific failures.
• Architect how GPU capacity is exposed to compute platforms and integrated with Kubernetes for GPUs and AI workloads.
• Drive reliability and performance at fleet scale with automated diagnosis and repair of unhealthy accelerators.
• Evaluate and onboard new GPU/AI accelerator platforms, networking topologies, and multi-node training/inference patterns.
• Establish standards, tooling, and APIs for safe, efficient GPU compute usage across teams.

Skills

Go programming

Distributed systems

GPU compute

Tools

Kubernetes

CUDA

NVLink

InfiniBand

RoCE

Bare-metal concepts

BMC/IPMI/Redfish

Job description

Roblox is seeking a Principal Software Engineer, GPU Compute, in San Mateo, CA to lead GPU compute initiatives and own the machine management layer for production-scale accelerators. You will partner with Kubernetes, networking, and cloud teams to drive end-to-end GPU strategy and reliability.

The role requires deep GPU expertise, strong Go skills, and experience operating GPU/AI workloads in production. Onsite work in our San Mateo office is expected with equity and benefits available.