AI Network Engineer

AMSYS Innovative Solutions, LLC; Houston (TX); On-site; USD 150,000 - 190,000; Full time

Job summary

This hands-on, customer-facing role requires collaboration with product engineering and field teams to translate designs into field-ready solutions and deliver scalable, high-density deployments.

Qualifications

• 10+ years designing and deploying network infrastructure for AI, HPC, or large GPU clusters.
• Deep hands-on InfiniBand experience: rail-optimized topologies, Adaptive Routing, SHARP, congestion control.
• RoCEv2/RDMA tuning at scale, plus strong BGP, EVPN/VXLAN, and multi-tenant design.
• Experience with BlueField DPUs and DOCA.
• A track record of writing architecture docs and runbooks that engineers actually use.

Responsibilities

• Design reference architectures for AI training, inference, and HPC environments.
• Create deployment runbooks and performance baselines for scalable GPU clusters.
• Collaborate with field teams to translate designs into field-ready solutions.
• Build customer-facing documentation and playbooks to support delivery and pre-sales.

Skills

AI network design

GPU cluster networking

InfiniBand

BGP

EVPN/VXLAN

ROCEv2/RDMA tuning

BlueField DPUs

Architecture docs & runbooks

Tools

Ansible

Terraform

Python

Job description

We are looking for a Sr AI Network Engineer to design and validate the fabrics behind large-scale GPU clusters for a global technology leader's AI and Hybrid Cloud engineering team. You'll build reference architectures, deployment runbooks, and performance baselines for AI training, inference, and HPC environments using NVIDIA Spectrum-X Ethernet, Quantum InfiniBand, and BlueField-3 DPUs. Your designs become the playbook that pre-sales, professional services, and delivery teams take into the field.

This is a hands-on, customer-facing role, working closely with product engineering on high-density, liquid-cooled rack deployments.

What you bring

• 10+ years designing and deploying network infrastructure for AI, HPC, or large GPU clusters
• Deep hands-on InfiniBand experience: rail-optimized topologies, Adaptive Routing, SHARP, and congestion control
• RoCEv2/RDMA tuning at scale, plus strong BGP, EVPN/VXLAN, and multi-tenant design
• Experience with BlueField DPUs and DOCA
• A track record of writing architecture docs and runbooks that engineers actually use

Nice to have

• NVIDIA networking or InfiniBand certification, or CCIE/CCNP Data Center
• Automation with Ansible, Terraform, or Python
• NeoCloud or managed service provider background
• Exposure to FedRAMP or SOC 2 environments