GPU & ML Infrastructure Engineer(Exp : 10+ yrs)

Experience: 10+ years preferred (8+ years for exceptional candidates) Location: Remote Overlap: Must be available until 11:00 AM PST Role : We’re looking for an experienced GPU & ML Infrastructure Engineer to build and maintain infrastructure for NVIDIA GPU benchmarking, stress testing, ML workloads, telemetry, and data generation across data-center and edge/embedded platforms. Responsibilities

• Deploy and optimize LLM inference/training workloads, including Llama-family models.
• Build GPU benchmarks and synthetic workloads: GEMM, convolution, compute, memory bandwidth, multi-GPU, and vision.
• Collect and validate GPU/hardware telemetry using NVML, DCGM, BMC, IPMI, and Redfish.
• Automate infrastructure deployment, workload execution, logging, data collection, and validation.
• Diagnose GPU sensor, sampling, timing, power, and thermal issues.
• Ensure workloads are reproducible and telemetry/data is accurate and well documented.
• Maintain hardware, driver, firmware, workload, and run metadata.

Required Skills

• Strong experience with NVIDIA GPUs and Linux systems.
• Hands-on LLM inference/training experience.
• Experience with quantized and full-precision models.
• Strong knowledge of GPU drivers, process orchestration, scheduling, and automation.
• Experience collecting and analyzing hardware telemetry programmatically.
• Understanding of workload reproducibility and performance benchmarking.
• Strong scripting/automation skills.

Nice to Have

• NCCL, multi-GPU scaling, tensor/pipeline parallelism
• GPU benchmarking/stress-testing
• GPU power/thermal management and throttling
• Hardware qualification or performance testing
• Time-series telemetry and data-quality validation

Ideal candidate: An independent engineer comfortable owning the complete workflow—from bringing up new GPU platforms and running ML workloads to collecting reliable telemetry and producing reproducible datasets.

Pay: ₹150,000.00 - ₹175,000.00 per month

Benefits:

• Provident Fund

Work Location: Remote