Amundson Group
The Role
Our client is a self-performing contractor building standardized modular units across multiple fabrication yards and deploying them as plug-and-play sites nationwide. They run 20 to 30 sites in development at any given time, and they need a Network Architecture Engineer who owns the entire network fabric connecting them. This is hands-on infrastructure work for AI and GPU workloads at scale, not office networking.
What You Will Run
You will design and build data center networks from the ground up. That means both the physical plant and the logical architecture working together as one system.
- Data center topology: leaf-spine fabrics, traffic patterns between racks, rows, and halls, redundancy and failover design.
- Rack-level design: top-of-rack versus end-of-row decisions, cabling, patching, power and network layout inside the rack.
- Fiber plant and optics: cable types, connectors (MPO/MTP), optics selection (400G, 800G, coherent), and long-haul circuits.
- Routing and overlay: BGP underlay, EVPN-VXLAN, and congestion management for GPU clusters (PFC, ECN, RoCE).
- Field work and 24×7 on-call support for a growing network of production sites.
What It Takes
You have physically built and logically designed data center networks. You know what an IT rack looks like, how it is wired, and how to connect hundreds of them into a resilient, high-performance fabric.
- Must-have: hands-on experience designing or building leaf-spine fabrics. You can describe spine and leaf counts, oversubscription ratios, and how racks connect to the fabric.
- Must-have: comfort with ToR versus EoR, fiber cable plant design, and optics selection for data center environments.
- Must-have: field work in data centers, colos, or internet exchanges. You understand rack power density, cooling, and the physical constraints that drive design.
- Must-have: BGP, EVPN-VXLAN, or equivalent fabric routing protocols.
- Strong plus: GPU cluster or AI infrastructure networking experience, including RoCE, InfiniBand, or rail-optimized designs.
- Strong plus: automation and scripting (any language).
- Strong plus: long-haul or DCI circuit design and carrier handoff experience.
Red flags: experience limited to office LAN, Wi-Fi, firewalls, or Meraki. Certifications without data center build work to back them up. Configuration skills without fabric design or cable plant experience.
Comp and Benefits
175,000 to 200,000 per year.
The Work
You will be part of a founding team with deep energy and infrastructure experience. The role includes 24×7 on-call rotation and 10 to 30 percent travel to sites. Candidates should be based in Texas or willing to relocate, and should be ready to start soon.
At a glance
- Location: Texas
- Compensation: $175,000 to $200,000 per year
- Recruiter: Amundson Group, veteran-owned, construction-only
Apply or ask about this role. Conversations are confidential; if this role is not the fit, we likely have the one that is.