Infrastructure SRE - HPC
Core
Operate a large, multi-vendor GPU fleet supporting both long-running distributed training jobs and low-latency inference services, solving reliability issues in storage, networking, and GPU systems.
Role type
Senior Infrastructure SRE (HPC)
Builds
Internal tooling for fleet observability, capacity management, and incident response.
Domain
High-performance computing, AI infrastructure, GPU clusters
Deliverable
infrastructure
Required skills
Python or Go, parallel filesystems (Lustre, GPFS, WEKA, BeeGFS), RDMA networking (InfiniBand, RoCE), GPU systems (NCCL, DCGM), Kubernetes/Slurm, distributed training reliability
Preferred skills
Slurm/Kubernetes hybrid environments, on-premise GPU deployment, Indian NCPs experience, multi-tenant GPU isolation
Technologies
Python, Go, Lustre, GPFS, WEKA, BeeGFS, InfiniBand, RoCE, NVLink, NVSwitch, NCCL, DCGM, Kubernetes, Slurm
Responsibilities
Operate GPU fleet end-to-end (provisioning, observability, capacity, health), manage on-call rotation and write runbooks, build internal tooling, partner with ML/platform teams to ensure workload stability
Seniority
Senior, hands-on IC