Jobs via Dice · 3 hours ago
Cloud Infrastructure Engineer (GPU/DPU Specialist)
United States
Full-time
Remote
Senior Level
$140K/yr - $150K/yrDice is the leading career destination for tech experts at every stage of their careers. Our client, GlobalLogic Inc., is seeking a Cloud Infrastructure Engineer specializing in GPU and DPU technologies to define end-to-end architecture for GPU compute and high-performance networking. The role involves designing lossless datacenter networking and driving architectural decisions across various technologies.
Computer Software
Responsibilities
Define E2E architecture for GPU compute and high-performance networking using hardware offload (NICs, DPUs, GPUs)
Design lossless, RDMA-enabled datacenter networking
Drive architectural decisions across SR-IOV, vDPA, RDMA/RoCEv2, VXLAN, and firewall placement
Align infrastructure design with Kubernetes/OpenStack and multi-tenant operational models
Act as the technical authority across networking, systems, virtualization, and GPU performance domains
Qualification
Required
Multi-GPU systems, GPUDirect RDMA, NCCL/MPI scaling, PCIe vs NVLink tradeoffs
Hardware offloaded networking (RDMA/RoCEv2, lossless Ethernet (PFC/ECN), VXLAN overlays, NIC/DPU offloads)
Linux networking and memory model, DMA/IOMMU, PCIe fabric, NUMA locality
Virtualization and cloud integration (KVM, GPU passthrough/vGPU, SR-IOV, vDPA, Kubernetes/OpenStack integration)
Performance and security tradeoffs (stateless vs stateful firewalling, datapath placement, observability strategy)
Define E2E architecture for GPU compute and high-performance networking using hardware offload (NICs, DPUs, GPUs)
Design lossless, RDMA-enabled datacenter networking
Drive architectural decisions across SR-IOV, vDPA, RDMA/RoCEv2, VXLAN, and firewall placement
Align infrastructure design with Kubernetes/OpenStack and multi-tenant operational models
Act as the technical authority across networking, systems, virtualization, and GPU performance domains
Bachelor's or Master's degree in Computer Science, Computer or Electrical Engineering, Mathematics, or a related field
Preferred
Be familiar with NVIDIA GPUs, ConnectX NICs, and BlueField DPUs
Experience with RDMA-enabled, lossless Ethernet fabrics for NVIDIA GPU clusters
Understanding of GPUDirect RDMA, NCCL scaling, and GPU interconnect topologies
Ability to align infrastructure architecture with NVIDIA reference designs and support models
Company
Jobs via Dice
Welcome to Jobs via Dice, the go-to destination for discovering the tech jobs you want.
0-1 employees
Funding
Current Stage
Early StageCompany data provided by crunchbase