Lead Architect (Network Infrastructure)

SBT; San Francisco (CA); Sur place; USD 210 000 - 320 000; Plein temps

Résumé du poste

SBT is partnering with a confidential AI startup to recruit a seasoned Network Architect for scale-out AI/GPU compute environments in San Francisco. The role leads end-to-end architecture for high-performance networking, shaping the network fabric strategy and interconnects across large clusters.

The candidate will collaborate across hardware, firmware, software, and systems teams, own deployments, and drive vendor evaluations to optimize platforms and performance at scale.

Qualifications

• 10+ years of experience designing and architecting high-performance networking infrastructure within AI, HPC, or hyperscale data center environments.
• Deep expertise in AI cluster networking technologies, including InfiniBand, RoCEv2, Ethernet fabrics, high-speed interconnects, and spine-leaf architectures.
• Proven hands-on ownership of large-scale AI/GPU clusters from architecture through deployment, including network design, implementation, performance tuning, and troubleshooting.
• Strong background in network and systems architecture, with the ability to influence hardware, firmware, software, and platform‑level product decisions.

Responsabilités

• Lead the end-to-end architecture, design, and deployment of high-performance networking infrastructure supporting large-scale AI/GPU compute clusters.
• Define the overall network fabric strategy, including next-generation interconnect architectures, scalability, performance optimization, and system-level integration.
• Collaborate with hardware, firmware, software, and systems engineering teams to influence product definition, platform architecture, and future technology roadmaps.
• Partner with technology vendors to evaluate solutions, drive technical tradeoffs, and co-develop optimized networking platforms.
• Take hands-on ownership of AI cluster networking from initial architecture through deployment, addressing performance bottlenecks, reliability challenges, and operational efficiency.

Connaissances

InfiniBand

RoCEv2

Spine-leaf

High-performance networking

Network architecture

Description du poste

SBT is the exclusive retained recruiting firm for this search.

This confidential client is an AI startup developing custom silicon and next-generation computing infrastructure for AI inference. The company is entering a significant growth phase, expanding its engineering capabilities and developing advanced platforms for large-scale deployments. They are seeking an experienced Network Architect to help define future product architectures, drive critical technical decisions, and deliver high-performance, scalable solutions.

Role

• Lead the end-to-end architecture, design, and deployment of high-performance networking infrastructure supporting large-scale AI/GPU compute clusters.
• Define the overall network fabric strategy, including next-generation interconnect architectures, scalability, performance optimization, and system-level integration.
• Collaborate with hardware, firmware, software, and systems engineering teams to influence product definition, platform architecture, and future technology roadmaps.
• Partner with technology vendors to evaluate solutions, drive technical tradeoffs, and co-develop optimized networking platforms.
• Take hands-on ownership of AI cluster networking from initial architecture through deployment, addressing performance bottlenecks, reliability challenges, and operational efficiency.

Qualifications

• 10+ years of experience designing and architecting high-performance networking infrastructure within AI, HPC, or hyperscale data center environments.
• Deep expertise in AI cluster networking technologies, including InfiniBand, RoCEv2, Ethernet fabrics, high-speed interconnects, and spine-leaf architectures.
• Proven hands-on ownership of large-scale AI/GPU clusters from architecture through deployment, including network design, implementation, performance tuning, and troubleshooting.
• Strong background in network and systems architecture, with the ability to influence hardware, firmware, software, and platform‑level product decisions.