Head of AI Infrastructure — Cluster SRE & Operations (Network)

Nava; Bengaluru; On-site; INR 3,500,000 - 7,000,000; Full time

Job summary

Nava is seeking a Network SRE & Operations lead to own production reliability of Nava's GPU networking fabric. You’ll drive proactive telemetry, incident response and continuous improvements across InfiniBand, RoCE and Ethernet fabrics in a high-performance AI data center environment.

You will lead a team of network SRE engineers, implement automation, and coordinate with platform and vendor teams to ensure lossless, scalable GPU workloads and NCCL performance targets are met.

Qualifications

• 10–15 years operating large production data center networks with team leadership.
• Deep BGP, EVPN/VXLAN, Linux networking and troubleshooting expertise.
• InfiniBand or RoCEv2, RDMA, PFC, ECN and congestion analysis skills.
• Python/Go/network automation, observability and incident leadership experience.
• Ability to correlate network failures with GPU distributed-workload behavior.
• Hands-on GPU/AI-datacenter exposure is valued; depth preferred over years.
• GPU/HPC fabric SRE, NCCL performance and advanced vendor telemetry.
• Background at hyperscalers or high-performance networking environments.

Responsibilities

• Own network availability, service indicators and error budgets across InfiniBand/RoCE, Ethernet and management fabrics.
• Lead severe incident diagnosis involving congestion, packet loss, RDMA failures, routing instability, link and optics faults.
• Build fabric telemetry, alerting, fault correlation and operational health dashboards.
• Automate configuration checks, safe changes, diagnostics, recovery and resilience testing.
• Coordinate network capacity and firmware lifecycle with deployment, platform and vendor teams.
• Lead network SRE engineers, incident reviews and prioritized reliability improvements.
• Correlate fabric telemetry (PFC pause, ECN marks, buffer occupancy, link flaps) with NCCL performance to detect issues before jobs degrade.
• Own lossless-RDMA health as an SLI — monitor PFC/ECN behavior and adaptive-routing balance via UFM (InfiniBand) and NetQ/Spectrum-X, driving recurring-incident elimination.

Skills

Network leadership

BGP EVPN

Linux networking

RDMA

Python

Observability

GPU datacenter

NCCL performance

Tools

InfiniBand

RoCEv2

NetQ

Spectrum-X

UFM

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 SRE & Operations team owns operational reliability of Nava's production GPU network fabrics, from proactive telemetry to rapid mitigation and the long-term elimination of recurring failures.
Own production reliability of GPU cluster networking, including InfiniBand, RoCE, and Ethernet fabrics. Lead network monitoring, incident response, performance troubleshooting, capacity planning, operational automation, and continuous availability improvements.

What You’ll Do

• Own network availability, service indicators and error budgets across InfiniBand/RoCE, Ethernet and management fabrics.
• Lead severe incident diagnosis involving congestion, packet loss, RDMA failures, routing instability, link and optics faults.
• Build fabric telemetry, alerting, fault correlation and operational health dashboards.
• Automate configuration checks, safe changes, diagnostics, recovery and resilience testing.
• Coordinate network capacity and firmware lifecycle with deployment, platform and vendor teams.
• Lead network SRE engineers, incident reviews and prioritized reliability improvements.
• Correlate fabric telemetry (PFC pause, ECN marks, buffer occupancy, link flaps) with NCCL collective performance to detect congestion and silent packet loss before jobs degrade.
• Own lossless-RDMA health as an SLI — monitor PFC/ECN behavior and adaptive-routing balance via UFM (InfiniBand) and NetQ / Spectrum-X, and drive recurring-incident elimination against NCCL bus-bandwidth and PFC-pause targets.

What You’ll Need

• 10–15 years operating large production data center networks with team leadership.
• Deep BGP, EVPN/VXLAN, Linux networking and troubleshooting expertise.
• InfiniBand or RoCEv2, RDMA, PFC, ECN and congestion analysis skills.
• Python/Go/network automation, observability and incident leadership experience.
• Ability to correlate network failures with GPU distributed-workload behavior.
• 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.
• GPU/HPC fabric SRE, NCCL performance and advanced vendor telemetry.
• Background at hyperscalers or high-performance networking environments.

Skills: infrastructure,reliability,infiniband,rdma,networking