AI Factory Network Architect

This is a fully remote position, open to applicants in United States, +1 more country.

📋 Description

• Take ownership of the complete network architecture for the AI Factory, encompassing NVLink/NVSwitch, InfiniBand/RoCE, storage, management, pod-to-pod, and inter-site networks.

• Design the optical layer, which includes transceivers, cabling, co-packaged and linear-drive optics, DWDM, link budgets, and reliability targets.

• Establish a repeatable reference network design for Sonic Inference Pods.

• Develop and initiate initial fabrics, incorporating rail-optimized topologies, subnet management, routing, adaptive routing, congestion control, QoS, and GPUDirect RDMA tuning.

• Ensure alignment of topology and buffer decisions with inference traffic patterns, such as collectives, KV-cache transfer, and multi-node serving.

• Oversee capacity planning, vendor roadmaps, technical specifications, and fabric-generation assessments.

• Implement telemetry, link-health monitoring, failure-domain design, automated fault isolation, and operational runbooks.

• Recruit and manage network engineers as the infrastructure footprint expands.

• Act as the final technical point of escalation for fabric incidents across all locations.

• Collaborate with hardware, platform, inference engineering, NVIDIA, optical and switching vendors, carriers, and site deployment teams.

⛳️ Requirements

• Over 15 years of experience in designing, building, and operating high-performance networks.

• Extensive experience in HPC or AI supercomputers at the scale of thousands of GPUs or more.

• Profound hands-on experience with InfiniBand, including fabric design, subnet management, routing engines, adaptive routing, congestion control, and port/queue-pair performance debugging.

• Familiarity with NVLink and NVSwitch domains and the interactions of scale-up/scale-out fabrics.

• Experience in optical networking with 400G and 800G transceivers and cabling, link budgets, DWDM, and coherent optics.

• Practical RDMA experience across InfiniBand and RoCE, including GPUDirect, NCCL, collective behavior, PFC, and ECN tuning.

• Proven track record of leading network teams through greenfield projects and production incidents.

• Experience with infrastructure automation using Python or Go, infrastructure as code, and telemetry pipelines.

• Strong technical writing abilities.

• Willingness to travel to pod sites located in the US and Europe.

• Key differentiators include experience with NVIDIA Spectrum-X, Quantum-X800, Blackwell-generation NVLink systems, co-packaged optics, linear-drive pluggables, photonic switching, modular/containerized data centers, inference-specific traffic patterns, SONiC, OpenFabrics, UEC, sovereign/data-residency network designs, as well as GPU cloud, hyperscaler, national lab, or AI lab environments.

🏝️ Benefits

• Remote-first work arrangement.

• Company retreats held in-person twice a year.

• Flexible core hours and autonomy over scheduling outside of collaboration blocks.

• Real downtime is provided following major release cycles.

• Generous paid time off, including vacation, sick days, and public holidays.

• Meaningful stock options available.

• Paid maternity, paternity, and caregiver leave.