GPU & AI Infrastructure Architect (Data Center)

Roblox; San Mateo (CA); On-site; USD 243,000 - 295,000; Full time

Job summary

Roblox is seeking a software-hardware specialist for the Infrastructure Foundation Hardware Engineering team in San Mateo, CA. You will own GPU hardware lifecycle from architectural evaluation to fleet integration and performance tuning.

You will prototype GPU platforms, lead firmware qualification, and collaborate with vendors to resolve complex issues, while building advanced GPU monitoring for real-time rendering and AI workloads.

Qualifications

• BA/BS in Electrical Engineering, Computer Engineering, or related field.
• 5 years of hardware engineering experience with a focus on GPU architecture and HPC.
• Deep technical knowledge of data center tech: PCIe, NVLink, InfiniBand, and liquid cooling.
• Hands-on testing of CPU, Memory, NVMe, and networking subsystems in Linux.
• Programming in Python, Go, or C for hardware validation tools.
• Strong debugging skills, including kernel logs and hardware registers.

Responsibilities

• Prototype next-generation GPU-accelerated hardware platforms and ensure integration with high-density compute nodes.
• Drive GPU integration, performance testing, and hardware-level optimizations.
• Develop evaluation and stress-testing strategies for GPU-heavy server platforms.
• Lead firmware qualification (BIOS/BMC) and automation for GPU health.
• Collaborate with hardware vendors and triage failures across hardware, driver, and kernel layers.
• Build and maintain monitoring stacks (Grafana/Prometheus) to track GPU metrics.

Skills

GPU architecture

Python

Go

C

Education

BA/BS in Electrical Engineering or Computer Engineering

Tools

Linux

BIOS/BMC

Job description

Roblox is seeking a software-hardware specialist for the Infrastructure Foundation Hardware Engineering team in San Mateo, CA. You will own GPU hardware lifecycle from architectural evaluation to fleet integration and performance tuning.

You will prototype GPU platforms, lead firmware qualification, and collaborate with vendors to resolve complex issues, while building advanced GPU monitoring for real-time rendering and AI workloads.