SoC Product Architect, Memory Ecosystem

NVIDIA•Us, CA

17h

About The Position

NVIDIA’s SoC Product Architecture team defines the platforms that power AI, graphics, and accelerated computing across data center, automotive, client, and edge products. As a Senior SoC Product Architect, Memory Ecosystem, you will drive NVIDIA’s external memory ecosystem across DRAM and non-volatile flash, working with major memory vendors to align roadmaps, co-innovate new technologies, and ensure our SoC portfolio delivers industry-leading bandwidth, latency, power, and cost.

Requirements

MS in Electrical Engineering, Computer Engineering, Physics, or related field, or equivalent experience.
18+ years of experience in DRAM / memory subsystem architecture, SoC or memory controller design, or closely related roles in semiconductor or system companies.
Deep understanding of modern DRAM technologies (e.g., DDR, LPDDR, GDDR, HBM) and non-volatile memories (e.g., NAND flash, managed NAND/SSD), including key tradeoffs in bandwidth, latency, power, capacity, packaging, and reliability.
Proven experience working directly with memory vendors on product definition, bring-up, or co-development, and familiarity with how large-scale customers influence vendor roadmaps.
Solid background in SoC and system memory subsystems, including memory controllers, PHYs, signal integrity, power delivery, and how memory choices impact overall SoC and platform performance and power.
Demonstrated ability to perform competitive analysis, including interpreting teardown data, benchmarking memory bandwidth/latency, and drawing clear architectural conclusions and recommendations.
Strong communication and presentation skills, with experience representing technical positions in vendor meetings, technical reviews, and multi-functional internal reviews.
Ability to work across functions (architecture, design, systems, sourcing, operations) and regions, driving alignment and decisions without direct authority.

Nice To Haves

Direct experience defining or architecting HBM, GDDR, or high-speed LPDDR solutions for GPUs, AI accelerators, or high-performance SoCs.
Hands-on work with advanced packaging (e.g., 2.5D/3D, CoWoS, multi-chip modules) involving high-bandwidth memory interfaces.
Prior ownership of memory-related driven teardowns or benchmarking programs, with clear examples of how results influenced product or technology direction.
Established relationships with major memory vendors and a history of successful roadmap or co-development engagements.

Responsibilities

Partner with all major memory vendors to understand their technology and product roadmaps (e.g., DDR/LPDDR, GDDR, HBM, NAND, emerging NVM) and align them with NVIDIA’s SoC and platform roadmaps.
Steer vendor roadmaps by articulating NVIDIA’s performance, power, capacity, form-factor, and reliability requirements, translating SoC and system needs into clear, data-driven feedback and requests.
Innovate new DRAM solutions in collaboration with vendors, exploring novel architectures, signaling, packaging, and power-management techniques to meet the bandwidth, latency, and power requirements of future NVIDIA SoCs and systems.
Work closely with SoC architects, memory controller designers, and board/system teams to co-optimize SoC + memory subsystem architecture, including channel topologies, speed grades, rank/stack configurations, and error-management strategies.
Represent NVIDIA at regular vendor technical reviews and key technical/roadmap meetings, presenting consolidated SoC roadmap needs, performance data, and proposals, and driving clear follow-ups and decisions.
Collaborate with sourcing, operations, substrate and PCB design, and quality teams to factor supply, cost, and reliability considerations into memory technology choices and long-term ecosystem strategy.