Senior Bare-Metal GPU Infrastructure Engineer

Primeintellect; San Francisco (CA); On-site; USD 150,000 - 300,000; Full time

Job summary

Prime Intellect is building the open superintelligence stack and a full-stack platform for frontier AI workloads. We seek an engineer to own the lifecycle of bare-metal GPU servers, from discovery to secure reuse, reducing manual work as the fleet scales.

You will automate provisioning, BIOS, and firmware, build hardware inventory services, and create robust tests before production. Collaborate across hardware, networking, and platform teams to deliver reliable infrastructure.

Qualifications

• 3+ years operating Linux servers or building bare-metal automation in production.
• Hands-on with PXE/iPXE, DHCP, image provisioning, and out-of-band management (Redfish or IPMI).
• Strong programming skills in Python or Go, plus Bash.
• Experience designing reliable automation that handles partial failures and drift.
• Ability to own operational incidents and collaborate across hardware, networking, and platform teams.

Responsibilities

• Build automated discovery, network boot, OS imaging, and configuration workflows for GPU servers
• Automate BIOS, BMC, NIC, GPU driver, and firmware configuration with staged rollouts and safe recovery paths
• Develop hardware inventory and lifecycle services that track machine identity, configuration, health, and readiness
• Create acceptance tests and burn-in workflows for GPUs, memory, storage, and interconnects before capacity enters production
• Integrate provisioning and health checks with SLURM, Kubernetes, and compute allocation systems
• Build observability, quarantine, repair, and re-provisioning workflows; partner with datacenter teams to resolve hardware failures
• Implement secure credential handling, tenant isolation, and data sanitization across the server lifecycle

Skills

Linux server ops

PXE/iPXE provisioning

Automation design

Python/Go/Bash

Incident ownership

Networking collaboration

Tools

Ansible

Terraform

MAAS/Ironic/Tinkerbell

Job description

Prime Intellect is building the open superintelligence stack and a full-stack platform for frontier AI workloads. We seek an engineer to own the lifecycle of bare-metal GPU servers, from discovery to secure reuse, reducing manual work as the fleet scales.

You will automate provisioning, BIOS, and firmware, build hardware inventory services, and create robust tests before production. Collaborate across hardware, networking, and platform teams to deliver reliable infrastructure.