Infrastructure/GPU Engineer

Cognizant is seeking a highly skilled hands-on Infrastructure Engineer with proven experience in the physical and technical deployment of AI-ready environments optimized for AI and machine learning workloads. This role focuses on NVIDIA DGX or similar systems, GPU-accelerated compute clusters, high-speed networking, and scalable storage solutions. The ideal candidate will have deep expertise in infrastructure design ,deployment, workload orchestration, and performance optimization in enterprise environments.

This is a remote role in the US. Salary range for this role is between $99,000 and $116,000 depending on skills and qualifications of the candidate. Applications will be accepted till 11/06/2025.

Key Responsibilities

System Design & Deployment

• Help in rightsizing GPU investment
• Architect and deploy NVIDIA DGX systems and GPU-based compute clusters.
• Design and implement scalable parallel filesystems (e.g., Lustre, BeeGFS, GPFS).
• Integrate high-speed interconnects using InfiniBand, RoCE, and RDMA.
• Collaborate on rack planning and airflow optimization.
  
  

Cluster & Infrastructure Management

• Configure and manage Slurm Workload Manager for job scheduling.
• Deploy and maintain cluster orchestration tools
• Automate provisioning using PXE boot, Terraform, Redfish, and Kubernetes.
• Perform firmware updates, BIOS/IPMI/BMC configuration, and OS provisioning
• Knowledge of Run.ai, ClearML or similar platform
  
  

Networking & Performance Optimization

• Design and validate network topologies including IPMI, internal/external networks, and InfiniBand fabrics.
• Optimize RDMA and RoCE configurations for low-latency, high-throughput data transfers.
• Conduct performance benchmarking using GPU-Burn, NCCL, and NVSM.
  
  

Monitoring & Troubleshooting

• Implement system health checks and diagnostics across compute, storage, and network layers.
• Troubleshoot hardware/software issues and ensure reliable infrastructure operation.
  
  

Required Skills & Qualifications

Technical Expertise

• Deep understanding of NVIDIA DGX architecture, CUDA, and GPU compute.
• Strong Linux system administration and shell scripting skills.
• Experience with Slurm, parallel filesystems, and high-speed networking (InfiniBand/RDMA/RoCE).
• Familiarity with containerization (Docker), orchestration (Kubernetes), and automation tools (Ansible, Redfish).
  
  

Preferred Qualifications

• Experience with BBCM, and DGX BasePOD/SuperPOD configuration
  
  

Certifications by Nvidia or equivalent OEM.