Staff GPU Memory Architect — High-Performance Mobile AI

Bzpp; United States; On-site; USD 151,000 - 252,000; Full time

Benefits offered by this job

Medical insurance

Dental insurance

Vision insurance

401(k)

Onsite lunch

Tuition assistance

Paid time off

Job summary

Samsung is recruiting a Staff GPU Memory Architect to define and implement next‑generation GPU memory subsystems and on‑chip interconnects for Exynos. In a mid‑to‑senior IC role, you will shape PPA across premium mobile GPUs, AI workloads, and ray tracing.

You’ll lead architecture studies, collaborate with GPU/CPU/NPU teams, and build performance models and validation plans. You will drive data‑driven decisions, develop architectural roadmaps, and champion innovative memory features to ensure

Qualifications

• 6+ years in GPU architecture, design, or development.
• Deep knowledge of memory hierarchies and on-chip interconnects.
• Experience with AI/ML workloads and ray tracing is a plus.
• Strong collaboration and communication skills.

Responsibilities

• Lead architecture studies for GPU memory subsystems and interconnects.
• Define features for next-gen memory subsystems to optimize PPA.
• Collaborate with cross-functional teams across SoC blocks.
• Produce roadmaps, design specs, and validation plans.

Skills

GPU architecture

High-performance design

Low-power design

Memory hierarchy

Performance modeling

Ray tracing / AI workloads

Education

Bachelor's in CS/Engineering

Master's degree

Ph.D. (optional)

Tools

RTL-level simulation

Analytical modeling

Job description

Samsung is recruiting a Staff GPU Memory Architect to define and implement next‑generation GPU memory subsystems and on‑chip interconnects for Exynos. In a mid‑to‑senior IC role, you will shape PPA across premium mobile GPUs, AI workloads, and ray tracing.

You’ll lead architecture studies, collaborate with GPU/CPU/NPU teams, and build performance models and validation plans. You will drive data‑driven decisions, develop architectural roadmaps, and champion innovative memory features to ensure