Senior Solutions Architect, Networking & GPU System

NVIDIA is a world‑leading, fast‑growing AI computing company, delivering everything from the most powerful GPU‑accelerated supercomputers to gigawatt‑scale AI data centers. We build and optimize end‑to‑end GPU server platforms that combine accelerated computing, high‑performance networking, and software to power state‑of‑the‑art AI infrastructure. We believe in our people and products, and we are looking for outstanding talent to join us. We are looking for a Senior Solutions Architect who is both customer‑facing and deeply hands‑on with SONiC‑based networking and GPU server infrastructure. In this role, you will partner with account and OEM partner teams to qualify opportunities, design solutions, run technical evaluations, and prove value to customers building large‑scale AI and HPC platforms. Supporting production clusters, including monitoring and troubleshooting, is also part of this role.

Work closely with account managers to understand customer requirements, position our SONiC + Spectrum whitebox switch solutions, and shape technical win strategies for AI and HPC opportunities. Design end‑to‑end architectures for customer proposals, including GPU server configurations, storage connectivity, and SONiC‑based leaf‑spine fabrics with RDMA/RoCE. Build and validate hands‑on demos and POCs: deploy SONiC switches and GPU servers, configure networking, install software stacks, and run benchmarks to prove performance, scalability, and reliability. Provide support for large-scale production SONiC switch clusters. Own and manage customized SONiC software projects on Spectrum white-box switches. Collaborate with internal engineering, product, and OEM partners to resolve complex issues in firmware, drivers, OS, routing, and GPU/network performance, then bring fixes back to active POCs. What we need to see BS/BA in Computer Science, Electrical/Computer Engineering, or equivalent practical experience. 6+ years in data center or cloud infrastructure roles (solutions architect, systems engineer, network engineer) with direct exposure to presales or customer‑facing technical work. Solid background in data center networking for AI workloads: leaf‑spine designs, high‑bandwidth/low‑latency fabrics, RDMA/RoCE, and ideally InfiniBand; comfortable configuring and debugging these in lab and customer environments. Practical knowledge of SONiC: installing and upgrading, configuring interfaces and routing (BGP/EVPN/VXLAN), using monitoring/telemetry tools, and troubleshooting real incidents. Strong, practical understanding of GPU server architecture: CPU/GPU balance, memory bandwidth, PCIe/NVLink topology, storage and NIC placement, and power/cooling at rack level. Hands‑on experience designing, deploying, or operating AI/HPC clusters using GPU‑accelerated servers (on‑prem or cloud), including real involvement in sizing, configuration, and performance tuning. Excellent communication and presentation skills: able to explain complex technical topics to both highly technical engineers and non‑technical decision makers, and write clear design documents and POC reports. Fluent written and spoken English is required. Ways to stand out from the crowd Contributions to open‑source SONiC, networking, or AI infrastructure projects, or published talks/whitepapers on AI data center design and performance tuning. Coding experience on SONiC. Coding experience with NCCL, NIXL or other collective communication library, RDMA applications, or performance‑critical distributed training frameworks, giving you credibility when discussing low‑level performance with customer engineers. Hands‑on deployments of SONiC in production or large lab environments, especially integrated with GPU clusters and RDMA/RoCE fabrics. Hands-on experience with NVIDIA networking products and solutions (Spectrum-X, InfiniBand, Cumulus Linux, etc.). With competitive salaries and a generous benefits