Network Architect: Data Center Fabrics for AI Training

Fluidstack; New York (NY); On-site; USD 150,000 - 210,000; Full time

Benefits offered by this job

Equity

Health, dental, and vision insurance

Pension/retirement plan

Generous PTO

Job summary

Fluidstack is seeking a Network/Infrastructure Architect to own the design of high-performance data center fabrics for AI workloads. You will craft lossless Ethernet fabrics, drive topology choices, and collaborate across hardware, DC Operations, ICT, and software to deliver scalable solutions.

You will translate requirements into deployable architectures for 100k+ accelerators, balancing power, cabling, and cooling while ensuring operational readiness and efficiency.

Qualifications

• Must have designed data center network fabrics from requirements through deployment.
• Deep L1–L3 expertise: CLOS/fat-tree topologies, BGP, EVPN/VXLAN.
• Experience designing lossless RDMA fabrics and understanding congestion at scale.
• Able to reason from first principles and collaborate across hardware, DC Ops, ICT, software.

Responsibilities

• Design topology designs, IP addressing, routing policy, and fabric configs for front-end, back-end, and storage networks.
• Translate logical designs into physical rack elevations, power constraints, cabling, and airflow considerations.
• Develop HLDs/LLDs, BOMs, cutsheets, and lead design reviews with stakeholders.
• Produce scalable designs adaptable to different GPU platforms and workload profiles.
• Work with cross-functional teams to ensure designs are buildable and operationally sound.

Skills

CLOS topology

BGP

EVPN/VXLAN

RDMA RoCEv2

Data center networks

Network design

Cross-functional collaboration

Performance tuning

Education

Bachelor's degree in CS/EE or related field

Job description

Fluidstack is seeking a Network/Infrastructure Architect to own the design of high-performance data center fabrics for AI workloads. You will craft lossless Ethernet fabrics, drive topology choices, and collaborate across hardware, DC Operations, ICT, and software to deliver scalable solutions.

You will translate requirements into deployable architectures for 100k+ accelerators, balancing power, cabling, and cooling while ensuring operational readiness and efficiency.