Roblox
Principal Software Engineer, GPU Compute
$345K–$399K
San Mateo, CA, United States · Principal · Full-time · Posted Jun 4, 2026
- Experience
- 10+ years building and operating large-scale distributed systems and infrastructure.
- Visa sponsorship
- For US-based roles, the company may not be able to employ candidates whose work authorization is related to certain U.S. visa categories, or support future H-1B sponsorship at this time.
About the job
Leads GPU compute strategy and builds reliable, production-ready GPU infrastructure for Roblox workloads across data centers and cloud environments.
What you'll do
- Own GPU host lifecycle, including driver and firmware management, GPU health telemetry, and failure remediation.
- Design GPU capacity exposure, scheduling, isolation, and Kubernetes integration for GPU and AI workloads.
- Drive fleet-scale reliability and performance; evaluate accelerator platforms and establish standards, tooling, and APIs.
What they ask for
- Deep hands-on GPU expertise in host provisioning, driver and firmware lifecycle, and production reliability.
- Strong proficiency in Go or other well-structured programming languages.
- Experience operating production GPU and AI workloads, including familiarity with CUDA, GPU scheduling, and high-performance networking.
- Track record scaling GPU or accelerator infrastructure used by other teams, with technical leadership experience.
Nice to have
- Familiarity with Kubernetes for GPU workloads.
- Familiarity with bare-metal concepts, including firmware, BMC/IPMI/Redfish, and OS imaging.
Pay
$345,040 — $399,420 USD (base salary)
Tools
Go, CUDA, Kubernetes, NVLink, InfiniBand, RoCE, BMC/IPMI/Redfish
AI focus
Owns infrastructure for GPU and AI accelerator workloads, including scheduling and isolation, platform evaluation, and multi-node training and inference patterns.
Model-written summary · the original posting is authoritative