The Role
Accountable for objective proof that the infrastructure is ready for production. Owns validation, performance baselines, defect closure, and the acceptance evidence that supports customer sign-off and go/no-go decisions.
Key responsibilities
• Create a requirements-to-test traceability matrix and staged entry/exit gates for site readiness, physical installation, power-on, fabric, storage, platform, burn-in, performance, resiliency, and handover.
• Develop version-controlled provisioning, health-check, diagnostic, stress, regression, and evidence-collection automation.
• Validate GPU or accelerator health, PCIe, GPU interconnect (for example NVLink/NVSwitch), NIC/HCA/DPU, InfiniBand or Ethernet/RoCE fabrics, storage throughput, Kubernetes, operators, scheduling, and telemetry.
• Execute failure and recovery tests including node loss, link degradation, service restart, workload rescheduling, rebuild, replacement, and return to service.
• Establish workload-relevant baselines for collective communications, network, storage, inference, training, and platform operations without assuming ownership of every customer model benchmark.
• Triage failures, isolate defects, coordinate corrective actions, retest, and maintain auditable defect and acceptance records.
• Produce formal acceptance packages containing configuration baselines, test results, exceptions, waivers, evidence, sign-offs, and residual risks.
• Use deployment findings and production incidents to extend automated coverage and prevent recurrence.
Kyndryl currently does not require employees to be fully vaccinated against COVID-19, however, if you are hired to work at a client, customer, or partner location, you may be required to show proof of vaccination to align with their respective COVID-19 vaccination policies. Those who believe they are eligible may apply for a medical or religious accommodation prior to the start of employment.
Who You Are
Required Experience
• 5+ years in infrastructure validation, automation, performance engineering, commissioning, or quality engineering.
• Strong Linux scripting and automation with Python, Ansible, APIs, Git, and CI/CD.
• Experience defining acceptance criteria, analyzing diagnostics, and managing defects to closure.
Preferred Skills
• NVIDIA DCGM diagnostics and telemetry, CVT, MLXlink, fabric validation, GPU burn-in, NCCL tests.
• Validation of DGX/HGX, InfiniBand, Spectrum-X, BlueField, Kubernetes operators, NVAIE, or Run:ai.
• Experience producing customer acceptance packages for NVIDIA DGX/HGX or NVIDIA-Certified AI infrastructure.
• NVIDIA or HPC benchmarking credentials and test-engineering credentials (for example ISTQB or equivalent).
Travel Requirements
Nationwide travel to customer data centers as required; percentage to be confirmed.
The compensation range for the position in the U.S. is - $143,640 to $273,000 based on a full-time schedule. Your actual compensation may vary depending on your geography, job-related skills and experience. For part time roles, the compensation will be adjusted appropriately. The pay or salary range will not be below any applicable state, city or local minimum wage requirement. There is a different applicable compensation range for the following work locations:
California (San Francisco Bay Area): $172,440 to $327,600
California (All Other): $158,040 to $300,360
Colorado: $143,640 to $273,000
Massachusetts: $143,640 to $300,360
New York City: $172,440 to $327,600
Washington: $158,040 to $300,360
Washington DC: $158,040 to $300,360
This position will be eligible for Kyndryl's discretionary annual bonus program, based on performance and subject to the terms of Kyndryl's applicable plans. You may also receive a comprehensive benefits package which includes medical and dental coverage, disability, retirement benefits, paid leave, and paid time off.
Note: If this is a sales commission eligible role, you will be eligible to participate in a sales commission plan in lieu of the annual discretionary bonus program.
Applications will be accepted on a rolling basis.
