Head of AI Infrastructure — Cluster Deployment (Network)

Nava; Bengaluru; On-site; INR 4,500,000 - 7,500,000; Full time

Job summary

Nava is building GPU-driven AI infrastructure in Bengaluru, delivering inference-as-a-service for teams building next-gen AI products. The Network Deployment team leads engineering and commissioning of high-performance networks across Nava's GPU clusters, with a hands-on leadership approach where the fabric is treated as code.

You will lead deployment and commissioning of InfiniBand, RoCE, Ethernet fabrics, own topology, automation and validation, and drive production readiness across multi-site

Qualifications

• 10–15 years in data center network engineering with leadership experience.
• Expertise in BGP, EVPN/VXLAN, ECMP, leaf-spine, and Linux networking.
• Solid knowledge of InfiniBand and RoCEv2, RDMA, PFC/ECN.
• Hands-on experience with high-speed switches (100/200/400G+) and network automation.
• GPU/AI-datacenter exposure is highly valued.

Responsibilities

• Lead installation, configuration and production acceptance of InfiniBand, RoCE and high-speed Ethernet fabrics.
• Own topology implementation, switch configuration, cable/optics validation, routing and management connectivity.
• Validate RDMA fabric behavior, bandwidth, latency, congestion and distributed GPU communications.
• Build automated configuration, discovery, preflight and regression tests for repeatable multi-site network commissioning.
• Resolve cross-domain integration defects with compute/storage engineers and network OEMs.
• Manage network deployment engineers, milestones, standards, documentation and production handover.
• Tune PFC watermarks and ECN thresholds to keep RDMA lossless under all-reduce incast; validate rail-topologies and adaptive routing.
• Define acceptance gates against measurable targets using NCCL all-reduce and telemetry validation.

Job description

About Nava

Nava is a neocloud company purpose-built for the AI era. We design, deploy, and operate large-scale GPU infrastructure and deliver inference-as-a-service to teams building the next generation of AI products. Our platform runs on NVIDIA GPU systems, high-performance RDMA fabrics, and a fully automated, software-defined operations model. Every engineer at Nava works close to the metal, on infrastructure built to keep GPUs saturated and models serving.

About Nava

Nava is a neocloud company purpose-built for the AI era. We design, deploy, and operate large-scale GPU infrastructure and deliver inference-as-a-service to teams building the next generation of AI products. Our platform runs on NVIDIA GPU systems, high-performance RDMA fabrics, and a fully automated, software-defined operations model. Every engineer at Nava works close to the metal, on infrastructure built to keep GPUs saturated and models serving.

About The Role

The Network Deployment team leads engineering and commissioning of high-performance networks across Nava's GPU clusters, including GPU fabrics, Ethernet leaf-spine, storage and out-of-band management networks. This is a hands-on leadership team where the fabric is treated as code. Lead deployment and commissioning of high-performance GPU cluster networks, including InfiniBand, RoCE, Ethernet fabrics, and management networks. Own network automation, integration, performance validation, troubleshooting, and production readiness.

What You’ll Do

• Lead installation, configuration and production acceptance of InfiniBand, RoCE and high-speed Ethernet fabrics.
• Own topology implementation, switch configuration, cable/optics validation, routing and management connectivity.
• Validate RDMA fabric behavior, bandwidth, latency, congestion and distributed GPU communications.
• Build automated configuration, discovery, preflight and regression tests for repeatable multi-site network commissioning.
• Resolve cross-domain integration defects with compute/storage engineers and network OEMs.
• Manage network deployment engineers, technical milestones, standards, documentation and production handover.
• Tune PFC watermarks and ECN thresholds to keep RDMA lossless under all-reduce incast; validate rail-optimized topologies and adaptive routing on the GPU backend fabric.
• Define acceptance gates against measurable targets — NCCL all-reduce bus bandwidth within target % of line rate and near-zero PFC pause — using UFM (InfiniBand) and NetQ / Spectrum-X telemetry, plus GPUDirect RDMA and BlueField DPU offload validation.

What You’ll Need

• 10–15 years in data center network engineering including leading network deployment teams.
• Advanced BGP, EVPN/VXLAN, ECMP, leaf-spine architecture and Linux network troubleshooting.
• Strong understanding of InfiniBand and/or RoCEv2, RDMA, PFC/ECN and fabric congestion.
• Hands-on high-speed switches (100/200/400G+) and Python/Ansible or equivalent network automation.
• Experience with network validation and root-cause analysis at hardware and protocol layers.
• Within that, hands-on GPU / AI-datacenter exposure is what we weigh most — we value demonstrated depth over years, as the field is still emerging.
• NVIDIA networking, NCCL/fabric benchmarking and large AI/HPC cluster builds.

Skills: validation,cluster,rdma,automation,network deployment