Overview
The HPC Network Engineer will take ownership of network operations within a dynamic tech-led business that manages a multi-tenant GPU cluster. This role will encompass end-to-end responsibilities, from network design through to day-to-day operations, requiring collaboration with other technical teams to ensure efficient network performance and tenant isolation.
Responsibilities
- Build and tune lossless RDMA fabrics for GPU and storage traffic.
- Run leaf-spine architecture with BGP and EVPN/VXLAN.
- Maintain tenant isolation across compute, storage, and management services.
- Implement network as code principles using Python, Ansible, and NetBox.
- Develop telemetry to identify congestion before it affects users.
- Act as the in-house network expert, including support for office network operations.
Requirements
- Minimum of 5 years experience with production data center networks.
- Strong expertise in BGP, EVPN/VXLAN, and Clos fabric configurations.
- Hands-on experience with RoCEv2 or InfiniBand technologies.
- Proficient in Linux environments for host-side debugging.
- Experience in GPU/HPC environments is essential.
- Familiarity with Spectrum-X or Quantum IB, ConnectX/BlueField, and NVMe-oF is a plus.