Principal Network Architect, RDMA & Datacenter
Overview
You own the datacenter scale-out architecture for the T100 platform, from the Linux host stack through the EIC PCIe bridge to third-party NICs and DMA into HBM. You define the transport and offload architecture, including RoCEv2 and Ultra Ethernet Transport (UET). You drive vendor selection and standard-setting in the networking community. Your work feeds a novel accelerator at scale, delivering up to 100 times the energy efficiency for AI inference.
What You'll Do9
- 1Own the end-to-end scale-out data path spanning host network stack, EIC PCIe bridge, NIC attachment, and DMA into HBM.
- 2Define the transport and offload architecture for RoCEv2 and Ultra Ethernet Transport (UET).
- 3Specify peer-to-peer DMA topology and access control services (ACS) policy across the PCIe hierarchy.
- 4Define queue and completion semantics exposed to the runtime, partnering with the software team.
- 5Drive NIC vendor selection and own technical engagement with networking silicon partners.
- 6Model and evaluate collective communication performance for distributed inference workloads.
- 7Own the architectural case and roadmap for a future Neurophos DPU and scale-up fabric.
- 8Represent Neurophos in standards bodies and the data center networking community.
- 9Establish networking architecture practices and mentor engineers across Architecture and Software.
Requirements10
- 110+ years in datacenter networking architecture, including RDMA transports deployed at scale.
- 2Deep knowledge of the Linux host networking stack: kernel networking, VFIO, DPDK or libfabric, and IOMMU/IOVA behavior.
- 3PCIe fluency: root complex versus endpoint roles, bridges, peer-to-peer transactions, ACS, and Gen5/Gen6 signaling.
- 4Familiarity with merchant NIC and switch silicon.
- 5Demonstrated ability to write architecture specifications that digital design and software teams implement against.
- 6Excellent communication skills and comfort leading cross-functional technical discussions.
- 7BS/MS/PhD in Computer Science, Electrical Engineering, or related field.
- 8Participation in Ultra Ethernet Consortium (UEC) or comparable standards body is preferred.
- 9Experience with DPU or SmartNIC architecture is preferred.
- 10Familiarity with collective communication libraries (NCCL, RCCL) is preferred.
Salary Insight
$250 - $300k per year
Similar open positions
Explore active roles that match your skills and interests.
Sciforium
VerifiedGPU Cluster Engineer Networking Sciforium
Senior Network Engineer leading GPU cluster networking at Sciforium. Own full stack from RDMA fabric to cloud connectivity. Design and operate high-performance networks for large-scale AI workloads. Differentiate by working directly with AMD engineers and scaling cutting-edge infrastructure.
Crusoe
VerifiedSenior Staff Software Engineer SDN Architecture
Crusoe seeks a Senior Staff Software Engineer to drive the software-defined networking stack enabling AI-first cloud infrastructure. This role involves designing high-performance networking solutions for massive GPU clusters while collaborating across product hardware and infrastructure teams.
Inabia Software & Consulting Inc.
VerifiedSenior Principal Hardware Engineer - Board & System Design
Senior Principal Engineer owns board and system hardware design for networking and AI infrastructure platforms. This role drives architecture from concept through production, supports customer engagements, and influences next-generation silicon. The engineer bridges silicon, hardware, and manufacturing teams with deep expertise in SerDes, PCB stack-up, and high-current power delivery. The role is a hands-on individual contributor with 15+ years of experience.
US63 NXP USA Inc.
VerifiedField Application Engineer – Robotics Networking
Lead technical leadership for robotics networking solutions driving scalable Ethernet and real-time communication for autonomous mobile robots drones and humanoid platforms. Architect in-robot network topologies and deliver proof-of-concept builds using NXP evaluation kits. Lead technical training and coordinate projects across customers and partners. Foster client-first mindset accelerating innovation in high-performance robotics solutions.
020 Cisco Systems, Inc.
VerifiedSenior Director Hyperscaler Systems Architecture
Lead definition strategy for Silicon Optics and Interconnects technologies focusing on Hyperscaler segment. Shape systems roadmap for scale-out scale-up and scale-across systems. Drive technical oversight through RFx approvals and simulations. Collaborate with customers partners and internal stakeholders to align architecture with business outcomes.
Staff Software Engineer Data Center Network Switches
Design and ship high‑performance networking solutions for Google’s global data centers. Lead the creation of scalable software that powers the world’s largest network infrastructure. This role drives innovation in AI/ML networks for TPUs and GPUs while collaborating across hardware and software domains. Join a diverse team to shape hyperscale computing and deliver exceptional performance.