Network Engineer

General Compute • San Francisco, CA

4d

About The Position

General Compute is a neocloud for alternative chips, productionizing purpose-built inference hardware for frontier labs, AI application companies, and asset-light clouds. The Network Engineer will design and build the network supporting the inference fleet, from racks and cabling to site connectivity and fabric performance. There is no existing network team, inherited topology, or reference architecture, so the engineer will establish a scalable network design that can be deployed across sites.

The fleet combines GPUs for prefill with decode accelerators from multiple ASIC vendors, each with distinct interconnect requirements and management tools. The role will define how these systems connect to each other, customers, and the internet, as part of the Data Center Deployment team responsible for bringing up each new site. The engineer will own the physical and Layer 2/Layer 3 network and work with the Platform Engineer, who owns fleet-wide request routing, model placement, and scheduling.

Requirements

• Hands-on experience designing and deploying data center networks, not only operating existing ones.
• Deep routing and switching knowledge for modern data center fabrics, including BGP, EVPN-VXLAN or equivalent, ECMP, and leaf-spine design.
• Experience building high-performance networks for AI/HPC clusters or other latency-critical environments, including relevant technologies such as RDMA/RoCE, InfiniBand, congestion control, and 400G/800G optics.
• Network automation skills using Python, Ansible, or similar, and comfort treating network configuration as code.
• Comfort working on the data center floor with cabling, optics, and hardware failures, directly or by guiding remote hands.
• Comfort with hardware heterogeneity and ambiguity.
• Genuine interest in being an early hire at a company of approximately 6–7 people.

Nice To Haves

• Networking experience with non-NVIDIA accelerators, such as SambaNova, Cerebras, or TPU pods.
• Multi-tenant cloud networking and tenant isolation at scale.
• Experience with transit contracts, peering, and internet exchanges.
• Experience with network source-of-truth tooling such as NetBox or Nautobot, and streaming telemetry such as gNMI.

Responsibilities

• Design and own the data center network architecture, including the leaf-spine fabric, high-bandwidth, low-latency paths between GPU prefill and ASIC decode systems, and out-of-band management.
• Lead network build-outs at new sites, including bills of materials, rack elevations, cabling plans, cross-connects, and bring-up through acceptance testing.
• Translate each accelerator vendor’s networking requirements into a consistent fleet design.
• Set up external connectivity, including IP transit, peering, and private customer connections.
• Build tenant isolation and network security into the network from the outset.
• Automate configuration, provisioning, and validation against a source of truth.
• Build network fabric monitoring and telemetry, and own network incident response.
• Manage network hardware vendors, carriers, colocation providers, and remote-hands teams.
• Own the physical and Layer 2/Layer 3 network, fabric performance, and site connectivity, handing off fleet-wide request routing, model placement, and scheduling to the Platform Engineer.

What This Job Offers

• Job Type: Full-time
• Education Level: No Education Listed