Senior Software Engineer - Cluster Networking

Overview As a Senior Networking Engineer, you will own the network architecture for NVIDIA's GPU superclusters, shaping how thousands of GPUs communicate across regions and clouds. You’ll lead Kubernetes networking at multi-thousand-node scale and design the overlay mesh, CNI data plane, and gateways that connect control and data planes. You’ll build scale-test environments and diagnostics to prevent networking regressions from impacting training workloads. This role sits at the core of our MARS initiative, enabling researchers to run frontier AI models at unprecedented scale. You will work with cross-functional teams to address hard, scale‑related networking challenges and drive impactful, reliable, Compensation / Benefitsequitybenefitsremote work optionscompetitive compensationopportunity to influence AI infrastructuremulti-cloud exposure ResponsibilitiesOwn and evolve Kubernetes networking architecture for GPU clusters at multi-thousand-node scaleDesign, operate and scale overlay network: CNI, mesh, VPN topologies, and gatewaysDesign, operate and scale L7 gateways/load balancers/tunnels (Envoy, Cloudflare)Identify scale ceilings: packet loss, control-plane saturation, IP exhaustion, and failure modes above thousands of nodesBuild scale-test environments and validation suites to catch networking regressionsDiagnose complex distributed network problems across stack and correlate root causesCollaborate with cloud providers on network topology and capabilities for new clustersProvide senior technical judgement to a distributed team in the Cluster Networking domain Key requirementsBS/MS in Computer Science, Electrical Engineering or related field, or equivalent experience6+ years in systems, network or infrastructure software engineeringDeep Kubernetes networking architecture knowledge with production Calico experience preferredProficiency with mesh and VPN topologies: Tailscale, WireGuard or equivalentStrong Linux networking fundamentals: routing, netfilter, iptables/nftables, packet marking, namespacesProven ability to debug distributed network problems at scale with packet capture and tracingProficiency in Go, Python, C or similar systems languageClear written and verbal communication across time zonesStrong communicationCross-functional collaborationStrategic thinkingKubernetes networking (CNI)Calico and CiliumTailscale