N

Nebius

Runs GPU clouds for AI training and deployment

GPU Cluster Architect

Cloud EngineeringConfirmed live in the last 24 hours
Full-timeRemote in USA

About the Job

Requirements

  • At least 5 years of experience designing clusters.
  • Deep understanding of modern GPU architectures, including NVIDIA and AMD.
  • Experience with high-performance computing interconnects, including InfiniBand and RoCE.
  • A solid background in systems architecture, networking, and hardware reliability.
  • Experience scripting for automation and telemetry pipelines using tools such as Python or Go.

Responsibilities

  • Architect scalable GPU cluster topologies, including compute nodes, interconnects such as InfiniBand and Ethernet, storage, and control planes.
  • Analyze AI and machine-learning workloads, including large language model training and inference, to inform design tradeoffs involving latency, bandwidth, and GPU density.
  • Align with network architects on relevant designs and validate low-latency, high-throughput interconnects, such as InfiniBand HDR/NDR and RoCEv2, at pod and data-center scale.
  • Work with storage teams to optimize performance for training datasets, checkpointing, and other workloads.
  • Understand and analyze signals from monitoring systems to detect design flaws.
  • Partner with site reliability, networking, storage, and data-center engineering teams to operationalize and scale the architecture.

About the Company

Nebius

Runs GPU clouds for AI training and deployment

Nebius runs an AI cloud for developers, researchers, and enterprises that train models and put them into production. Its AI Cloud combines GPU computing, storage, networking, and managed services, while Token Factory runs models through an API so teams do not manage the underlying servers. Customers rent computing and storage at usage rates or reserve capacity, choosing resources that fit short experiments or sustained production workloads. Beyond the cloud, the group owns Avride, which develops autonomous vehicles and delivery robots, and TripleTen, which teaches technical skills for people changing careers. The publicly listed group is headquartered in Amsterdam, Netherlands.

Company size
1,001-5,000
Company stage
IPO
Headquarters
Amsterdam, Netherlands
Founded
1997

Recently Posted Jobs

Explore more opportunities at Nebius

See more jobs