Senior System Architect

NeuReality is seeking a Lead System Architect to join our system architecture team and help define the next generation of our AI-SuperNIC scale-out chip.

AI scale-out communication is a critical element in modern data centers, and emerging standards such as Ultra Ethernet aim to address this challenge. This role focuses on defining a high-performance Smart NIC architecture optimized for GPU-centric AI workloads, with emphasis on low-latency, high-bandwidth data movement.

You will work across hardware and software domains, collaborating closely with AI, platform, driver, and VLSI teams to design a competitive scale-out networking solution.

Responsibilities

Define product requirements and architecture for NeuReality’s next-generation ultra low-latency AI-SuperNIC
Design end-to-end data paths across NIC, host, and GPU memory, with focus on GPU connectivity (PCIe, NVLink, CXL)
Lead adoption of emerging standards such as UEC and UALink, while maintaining compatibility with existing ecosystems
Establish architecture guidelines and partitioning across hardware, firmware, and software
Drive performance modeling, analysis, and system-level optimization for AI workloads
Optimize scale-out communication patterns (e.g., collectives) for GenAI training and inference
Collaborate with software teams on drivers, runtimes, and integration with AI frameworks
Work with R&D, CTO, and peer architects on system definition and trade-offs
Engage with customers and partners to shape system requirements and use cases

Requirements:

BSc/MSc in Electrical Engineering, Computer Science, or related field
5+ years of experience as a system architect on complex SoCs or data center systems
Strong experience with large-scale systems integrating hardware, firmware, and software
Background in high-performance systems, networking, or data center infrastructure
Experience working with GPU-based systems or AI/ML workloads
Knowledge of high-speed interconnects such as PCIe

Nice to have

Experience with UEC, UALink/NVLink, RoCEv2, NVMe-oF, or RDMA technologies
Familiarity with IP networking, TCP/UDP, or network security
Experience with in-network compute (e.g., SHARP, INC)
Deep understanding of GenAI/ML infrastructure and distributed workloads