Senior Solutions Architect – AI Infrastructure Networking

This is a fully remote position, open to applicants in United States.

📋 Description

• Take ownership of the network solution design for k0rdent AI, Mirantis's platform dedicated to constructing and managing GPU clouds and AI factories.

• Create interconnect designs for extensive GPU clusters, ensuring tenant isolation and workload connectivity across containers, virtual machines, and bare-metal nodes.

• Develop and publish network reference architectures and solution designs, ranging from single-rack setups to multi-thousand-GPU clusters.

• Define InfiniBand and RoCEv2 Ethernet compute fabrics, focusing on rail-optimized and fat-tree/Clos designs, as well as oversubscription and failure domains.

• Specify front-end, storage, and management networks, along with connectivity to customer data centers, public clouds, and hybrid environments.

• Detail multi-tenant isolation mechanisms utilizing InfiniBand partitions, VRFs, EVPN-VXLAN, and Kubernetes network policies.

• Document scalability limits and trade-offs concerning cost, performance, operability, and vendor lock-in.

• Define Linux host networking and facilitate the exposure of NICs, DPUs, and SuperNICs to workloads.

• Design networking for Kubernetes to support AI workloads, encompassing CNIs, Multus, SR-IOV, RDMA device plugins, NVIDIA Network Operator, and DRA.

• Create networking solutions for KubeVirt virtual machines, covering passthrough and SR-IOV for GPU and RDMA traffic.

• Investigate and assess emerging AI networking technologies and standards.

• Prototype alternative designs in the lab and compare them with current practices.

• Publish internal research notes, design proposals, and select external papers, blog posts, or presentations.

• Execute proofs of concept in collaboration with customers, partners, and on Mirantis or customer hardware.

• Develop automation and tooling using Python, Go, Bash, Ansible, Helm, Kubernetes manifests, and Terraform.

• Validate and benchmark fabrics and host configurations using tools such as NCCL tests, perftest, and ib_write_bw.

• Serve as the network subject matter expert during customer discovery, design reviews, and architecture workshops.

• Collaborate with hardware and networking partners on joint designs and validation efforts.

• Relay research findings to Product and Engineering teams to influence the k0rdent AI roadmap.

• Present at industry events, webinars, and partner summits.

• Conduct hands-on technical workshops and produce reference architectures, solution briefs, blog posts, and enablement content.

⛳️ Requirements

• A Bachelor's degree in Computer Science, Electrical Engineering, Telecommunications, or a related field, or equivalent practical experience.

• A minimum of 8 years in network engineering or network architecture.

• At least 3 years of experience in data center, HPC, or cloud infrastructure networking.

• Customer-facing experience in roles such as solutions architect, pre-sales engineer, consultant, or technical lead.

• Proficiency in data center network design, including spine-leaf and Clos topologies, rail-optimized GPU fabrics, oversubscription, and ECMP.

• Expertise in InfiniBand subnet management, partitioning, adaptive routing, and NCCL/RDMA traffic.

• Knowledge of RoCEv2 on Ethernet, covering PFC, ECN, DCQCN, QoS, MTU, and buffer tuning.

• Proficiency in BGP, EVPN-VXLAN, and VRFs for multi-tenancy.

• Experience with hybrid and multi-cloud connectivity, interconnects, VPN, cloud networking, IP addressing, and DNS planning.

• Solid understanding of Linux networking, including NIC drivers, PCI passthrough, SR-IOV, IOMMU, NUMA affinity, iproute2, ethtool, and devlink.

• Familiarity with Kubernetes networking, including CNI plugins, Multus, SR-IOV, RDMA device plugins, network policies, and service exposure.

• Experience with KubeVirt and VM networking, including passthrough and SR-IOV.

• Proficient in programming or scripting in at least one language, preferably Python or Go.

• Comfortable with Git, CI, and infrastructure-as-code practices.

• Excellent verbal and written communication skills in English.

• Capable of presenting to large audiences and conducting hands-on workshops.

• Able to work effectively across time zones and diverse work cultures.

• Remote candidates must reside in the United States East Coast.

• Willingness to attend some meetings outside of standard local hours.

• Preferred qualifications include experience with NVIDIA networking, GPU cloud/neocloud/HPC operators, bare-metal provisioning, network automation, SDN controllers, global load balancing, NVMe-oF, Mirantis products, open-source contributions, industry standards, published research, patents, or white papers, and proficiency in additional languages.

🏝️ Benefits

• Flexible remote work arrangement.

• Travel opportunities of up to 25% for customer engagements, partner meetings, lab work, and industry events.

• Collaborate with a world-class, distributed team.

• Opportunity to engage directly with leading GPU cloud operators, NeoClouds, sovereign clouds, and AI-first enterprises.

• Chance to influence product strategy and contribute to go-to-market success.