Staff Engineer, Bare-Metal & GPU Infrastructure

Mixpeek; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Modal is seeking engineers to design, build, and sustain the hardware fleet that powers its serverless platform. You will work on the machines layer, provisioning bare metal and cloud hosts, imaging, monitoring, and repairing them to keep production running smoothly.

You will automate capacity from various hardware providers, benchmark hosts, configure GPUs, RDMA, networking, and storage, and ensure rapid recovery from failures.

Qualifications

• 5+ years of experience writing high-quality production code.
• Experience operating fleets of physical hardware or building the related control planes.
• Strong cloud skills and Linux kernel/driver knowledge.
• Ability to debug across layers and respond to production incidents.
• Willingness to participate in on-call rotations.

Responsibilities

• Design, build, and maintain high-performance systems for Modal's serverless platform.
• Own the machines layer: provisioning, images, monitoring, and repairs of hardware fleets.
• Automate integration of new CPU/GPU/storage servers and hardware providers.
• Audit, benchmark, and maintain machine images, GPUs, RDMA, networking, and storage.
• Develop automation to keep the fleet healthy with minimal human intervention.
• Diagnose issues from kernel panics to container runtime on new architectures.

Skills

Cloud computing

Linux kernel & networking

Debugging across layers

On-call readiness

Go programming

Tools

BMC/IPMI

PXE/network boot

Bare metal provisioning

Kernel images & firmware

Job description

Modal is seeking engineers to design, build, and sustain the hardware fleet that powers its serverless platform. You will work on the machines layer, provisioning bare metal and cloud hosts, imaging, monitoring, and repairing them to keep production running smoothly.

You will automate capacity from various hardware providers, benchmark hosts, configure GPUs, RDMA, networking, and storage, and ensure rapid recovery from failures.