Summary
What you’ll impact
The Principal Network Engineer designs and evolves large‑scale RoCEv2 data‑center networks for AI/ML workloads. The role works with 400‑800 Gbps Ethernet fabrics, high‑speed optics, and switches from leading vendors, using Python, Ansible, and Terraform for automation and validation. Key duties include defining congestion‑management (PFC, ECN, DCQCN), building automation pipelines, and ensuring reliability across clusters of thousands to over 100,000 GPUs. The engineer partners with network architects and infrastructure leaders, sets technical direction across teams, and serves as the escalation point for complex scaling challenges.
Responsibilities
What you'll do
- As a Principal Engineer, you own end-to-end network architecture, make high-impact design decisions, and set technical direction across teams, with clear examples of systems you’ve defined and scaled
- Define, evolve, and standardize large-scale RoCEv2 data center networks supporting AI and ML clusters from thousands to 100,000+ GPUs
- Set and validate congestion management strategy across RDMA fabrics, including PFC, ECN, and DCQCN, based on real production behavior
- Establish automation, validation, and observability patterns that prevent misconfiguration and eliminate manual operational work
- Act as the technical escalation point for complex failures, scaling limits, and architectural tradeoffs in always-on, multi-tenant environments
Requirements
What you’ll bring
- Bachelor’s degree in Computer Science, Electrical Engineering, or a related technical field, or equivalent practical experience
- Deep experience designing and operating RDMA and RoCEv2 networks in large-scale production environments supporting AI or HPC workloads
- Expert-level knowledge of switching hardware and their NOS, such as Arista, Juniper, and custom solutions using SONiC, including high-speed Ethernet fabrics
- Proven hands-on experience with congestion management and performance tuning using PFC, ECN, and DCQCN
- Strong experience with high-speed optics and cabling including 400G, 800G, and AEC, AOC, DAC, and structured cabling at scale
- Strong automation mindset, with experience using Python, Ansible, Terraform, Git, and production observability tooling
- We’re looking for engineers who operate comfortably at scale, make hard calls with incomplete data, and take responsibility for systems that must work under sustained load. The solutions that work on a handful of devices will not work at Exascale.