Senior ML Infra Engineer: GPU Benchmarking & Cloud

Devconnectplatform; Amsterdam; Remote; EUR 90,000 - 130,000; Full time

Benefits offered by this job

Competitive pay

Career growth

Flexible work

International team

Job summary

Nebius is seeking a senior ML Infrastructure Engineer to lead benchmarking and optimization of GPU platforms for ML workloads. You will profile GPU performance, compare architectures, and develop tooling to visualize bottlenecks across CUDA and ROCm stacks.

Join an international, remote-first team building an AI cloud platform with strong emphasis on performance, scalability, and open-source benchmarking tools.

Qualifications

• Deep understanding of ML theory and foundations.
• Strong knowledge of performance aspects of large neural networks (training & inference).
• Experience with modern DL frameworks (PyTorch, JAX, Megatron-LM, Tensor-LLM).
• Solid grasp of GPU stack: CUDA, NCCL, drivers, libraries.
• Familiarity with containerized environments (Docker, Kubernetes).
• Excellent communication and independence.

Responsibilities

• Profile and analyse GPU performance at system/kernel level.
• Evaluate and compare GPU performance across platforms and stacks (CUDA, ROCm).
• Debug and optimise ML workloads on GPU, resolving bottlenecks.
• Perform acceptance testing for new GPU clusters ensuring performance and stability.
• Run experiments across diverse GPU configurations to study interconnects and system tuning.
• Develop tools and dashboards to visualise performance metrics and trends.
• Contribute to internal tooling, frameworks and best practices.

Skills

PyTorch

JAX

Megatron-LM

Tensor-LLM

CUDA

NCCL

Docker

Kubernetes

Python

AWS

GCP

Azure ML

Job description

Nebius is seeking a senior ML Infrastructure Engineer to lead benchmarking and optimization of GPU platforms for ML workloads. You will profile GPU performance, compare architectures, and develop tooling to visualize bottlenecks across CUDA and ROCm stacks.

Join an international, remote-first team building an AI cloud platform with strong emphasis on performance, scalability, and open-source benchmarking tools.