Senior GPU Cluster Engineer for AI Infrastructure

Sciforium; San Francisco (CA); On-site; USD 150,000 - 220,000; Full time

Benefits offered by this job

Medical, dental, and vision insurance

401k plan

Daily lunch, snacks, and beverages

Flexible time off

Competitive salary and equity

Job summary

Sciforium is seeking a GPU Cluster Engineer to own the software stack for high-performance GPU clusters. You will define production-ready node images, validate hardware with automated tests, and manage fleet upgrades while ensuring consistent, fast, and scalable performance.

You will work with two stakeholder groups—foundation model teams and model serving teams—deploying in a Kubernetes/Slurm environment with advanced NVIDIA/ROCm stacks and IaC tooling.

Qualifications

• Bachelor's or Master's in Computer Science, Computer Engineering, Electrical Engineering, or related field.
• 5+ years in systems/infrastructure with GPU cluster or ML infra experience.
• Deep Linux internals expertise: kernel modules, DKMS, systemd, cgroups, NUMA.
• Experience with NVIDIA CUDA and/or ROCm driver stacks on modern accelerators.
• Production Kubernetes experience with GPU workloads and HPC schedulers.

Responsibilities

• Own node software definition from base OS to production-ready GPU nodes.
• Build automated acceptance suites and validation checks for nodes.
• Maintain fleet with upgrades and topology-aware scheduling.
• Automate detection of unhealthy nodes and manage re-imaging workflows.

Skills

Linux internals

Python

Bash

GPU cluster engineering

NVIDIA CUDA

ROCm

Kubernetes

Slurm

Ansible/SaltStack

Education

Bachelor's or Master's in CS/CE/EE

Tools

Docker

NVIDIA Container Toolkit

Packer

MaaS

Foreman

Terraform

Job description

Sciforium is seeking a GPU Cluster Engineer to own the software stack for high-performance GPU clusters. You will define production-ready node images, validate hardware with automated tests, and manage fleet upgrades while ensuring consistent, fast, and scalable performance.

You will work with two stakeholder groups—foundation model teams and model serving teams—deploying in a Kubernetes/Slurm environment with advanced NVIDIA/ROCm stacks and IaC tooling.