GPU Cluster Site Reliability Engineer
Iframe
Operate GPU clusters hands-on with InfiniBand validation and Linux expertise across one of seven regions. This full-time individual contributor role keeps hardware and service reliability aligned.
As a Cluster Site Reliability Engineer at IC4–IC6, you will bring up GPU racks and maintain regional cluster operations. Your work spans NCCL or RCCL validation, Kubernetes at host level, Terraform, Prometheus, Grafana, and Go or Python tooling. The team deploys weekly and coordinates daily across seven regions.
Key Responsibilities:
- Bring up B200, B300, and MI300X GPU racks
- Validate NCCL or RCCL all-reduce performance
- Drive NDR or XDR InfiniBand fabric compliance
- Forecast regional capacity, power, and thermal headroom
- Respond to regional P1 incidents within stated targets
Requirements:
- Five-plus years operating large compute clusters
- Deep InfiniBand and Ethernet RoCE experience
- Subnet manager tuning and lossless networking expertise
- Go or Python tooling development experience
- Calm judgment during high-impact incidents
Apply GPU infrastructure, InfiniBand, and cluster reliability expertise across demanding regional operations and hardware bring-up work.
#J-18808-LjbffrReference: WJ-2742_652817 · Posted 20 Nov 2026, 11:35 PM
More jobs like this: