Principal Engineer, Rack-Scale GPU Architecture

This role owns the end-to-end reference architecture for rack-scale inference systems at DigitalOcean, covering GB300 NVL72 and future Vera Rubin systems. Responsibilities include defining NVLink domain partitioning, scale-out RDMA fabric design, making Kubernetes topology-aware, driving multi-node disaggregated serving, owning failure semantics and validation, and partnering with NVIDIA on pre-silicon enablement.

12+ years in large-scale compute, HPC, or GPU infrastructure engineering. Deep understanding of GPU interconnect and system topology (NVLink, NVSwitch, PCIe, NUMA). Strong RDMA networking background (RoCEv2 or InfiniBand at scale, GPUDirect RDMA). Practical NCCL expertise. Substantial Kubernetes depth for accelerated workloads. Strong systems programming in Go and Python. Experience operating fleets where hardware failure is routine. Excellent communication and cross-organizational leadership.