Specialist - Cloud Engineering

LTM; Chennai; In loco; INR 3.500.000 - 6.000.000; Tempo pieno

Descrizione del lavoro

NVIDIA in Chennai is seeking a GPU Platform Engineer DevOps to build and operate the GPU and inferenceserving stack for NVIDIA AI Enterprise on EKS. You will own GPU sharing scheduling, RunAI integration, and telemetry, collaborating closely with the architecture team to ensure high performance and reliability.

Responsibilities include deploying AI Enterprise on EKS, managing NIM microservices and Dynamo serving, tuning scheduling and autoscaling, and enabling efficient model onboarding and

Competenze

• 8 years in infrastructure ML engineering with hands-on NVIDIA GPU operations
• NVIDIA GPU stack drivers CUDA GPU Operator and DCGM
• DevOps Experience
• Kubernetes GPU workloads and device plugins
• GPU scheduling RunAI or equivalent and GPU partitioning MIG fractional GPU
• Autoscaling KEDA and OpenAI-compatible inference API patterns

Mansioni

• Deploy and operate NVIDIA AI Enterprise NVAIE on EKS: GPU Operator drivers CUDA runtime and DCGM
• Deploy and operate NIM microservices and NVIDIA Dynamo serving exposing the OpenAI-compatible API chat completions, embeddings, streaming, batch
• Configure GPU sharing MIG where SKU supports it and RunAI fractional GPU as isolation baseline
• Integrate RunAI for GPU scheduling quota and preemption and KEDA for replica autoscaling including scale-to-zero
• Instrument GPU telemetry via DCGM and drive inference performance validation and SLA-tier binding
• Support the model onboarding NIM packaging and inference-deploy pipelines
• Tune GPU serving performance and troubleshoot CUDA driver and scheduling issues

Conoscenze

NVIDIA GPU ops

GPU scheduling

DevOps

Kubernetes

Strumenti

CUDA runtime

DCGM

GPU Operator

Kubernetes

Descrizione del lavoro

The NVIDIA GPU Platform Engineer DevOps builds and operates the GPU and inferenceserving stack NVIDIA AI Enterprise on EKS NIM microservices and NVIDIA Dynamo serving and owns GPU sharing scheduling integration and serving performance Onshore placement supports handson GPUedge coordination and close work with the architecture team

Key Responsibilities

Deploy and operate NVIDIA AI Enterprise NVAIE on EKS GPU Operator drivers CUDA runtime and DCGM

Deploy and operate NIM microservices and NVIDIA Dynamo serving exposing the OpenAIcompatible API chatcompletions embeddings streaming batch

Configure GPU sharing MIG where the SKU supports it and RunAI fractional GPU as the isolation baseline and clearly distinguish hardware vs software isolation

Integrate RunAI for GPU scheduling quota and preemption and KEDA for replica autoscaling including scaletozero

Instrument GPU telemetry via DCGM and drive inference performance validation and SLAtier binding

Support the modelonboarding NIM packaging and inferencedeploy pipelines

Tune GPUserving performance and troubleshoot CUDA driver and scheduling issues

MustHave Skills Experience

8 years in infrastructureML engineering with handson NVIDIA GPU operations

NVIDIA GPU stack drivers CUDA GPU Operator and DCGM

DevOpd Experience

Kubernetes GPU workloads and device plugins

GPU scheduling RunAI or equivalent and GPU partitioning MIG fractional GPU

Autoscaling KEDA and OpenAIcompatible inference API patterns

NicetoHave Skills

NVAIE on AWSEKS and edge GPU Outposts G7

LLM RAG serving and vector databases

Inference performance benchmarking and optimization