Lead GPU Systems Architect for Tensor SoC

Socket.dev; Mountain View (CA); On-site; USD 163,000 - 236,000; Full time

Job summary

Google is seeking an experienced engineer to drive GPU architecture for the Tensor SoC, based on GPU workload analysis across high-end games, UI, and ML. You will propose system-level features to boost overall SoC performance and collaborate with product, research, and device teams to deliver compelling GPU-powered experiences.

The role requires deep knowledge of computer architecture and GPU pipelines, with emphasis on Vulkan/OpenGL/OpenCL and ARM-based systems, and will involve optimizing the

Qualifications

• Bachelor's degree or equivalent practical experience.
• 8 years of experience with computer architecture concepts, pipelining, or memory subsystems.
• Experience with system architecture or GPU workload analysis and optimization.
• Master's or PhD preferred with emphasis on computer architecture.

Responsibilities

• Drive GPU architecture for the Tensor SoC based on GPU workload analysis.
• Propose system level architectural features to improve SoC performance on GPU workloads.
• Collaborate with Product Management, Google Research, and device teams to bring GPU-powered experiences to Google.
• Work with GPU Software and Android teams to optimize the software stack for GPU workloads.

Skills

Computer architecture

Pipelining

Memory subsystems

GPU workload analysis

System architecture

GPU workload optimization

Education

Bachelor's degree

Master's or PhD in EE/CE/CS

Tools

Vulkan

OpenGL

OpenCL

Android OS

Firmware

ARM architecture

Job description

Google is seeking an experienced engineer to drive GPU architecture for the Tensor SoC, based on GPU workload analysis across high-end games, UI, and ML. You will propose system-level features to boost overall SoC performance and collaborate with product, research, and device teams to deliver compelling GPU-powered experiences.

The role requires deep knowledge of computer architecture and GPU pipelines, with emphasis on Vulkan/OpenGL/OpenCL and ARM-based systems, and will involve optimizing the