About Us

Vast.ai's cloud powers AI projects and businesses all over the world. We are democratizing and decentralizing AI computing — reshaping our future for the benefit of humanity. Our mission is to organize, optimize, and orient the world's computation.

We value elegance, ownership, integrity, and continuous learning. You'll have the opportunity to dive into state-of-the-art AI systems while collaborating with a globally distributed team.

About the Role

This is a systems operations support role focused on deep-diving into escalated infrastructure issues that go beyond frontline triage. You’ll be the engineering resource our L1 support team relies on when tickets become complex, investigating and resolving issues across the full infrastructure stack—including hardware, BIOS and firmware, networking, Ubuntu, Docker, NVIDIA CUDA and GPUs, and KVM virtual machines.

You’ll own complex escalations end-to-end: gathering evidence, reproducing issues, identifying the root cause, proposing solutions, and working with the appropriate teams to bring each issue to resolution. The best engineers in this role don’t just resolve individual tickets—they identify recurring patterns, improve operational tooling, and build runbooks that prevent future incidents. You’ll collaborate directly with the engineering and host support teams on systemic infrastructure issues.

Strong Linux systems knowledge, technical depth, and support experience are the primary requirements. You should be comfortable working autonomously in Ubuntu environments, troubleshooting hardware, networking, containers, virtual machines, and GPU workloads, and clearly communicating your findings and proposed solutions to both technical and non-technical audiences.

Vast.ai users or hosts strongly preferred.

Location and Schedule

This is a full-time position based in our Westwood, Los Angeles office.

Available schedules:

Monday–Friday: Fully on-site

Sunday–Thursday: Four days on-site and one day working from home

Key Responsibilities

Handle escalated support tickets involving GPU workload failures, container issues, networking problems, account infrastructure, and host-side configuration

Provide managed support for supplier onboarding and ongoing machine management, acting as a technical resource through installation, configuration, and post-setup troubleshooting

Assist clients and infrastructure suppliers working with TensorFlow, PyTorch, and other GPU-accelerated workloads

Provide coverage for L1 support overflow during peak periods or incidents

Diagnose and resolve issues across Docker, NVIDIA CUDA/GPU drivers, and KVM virtualization environments

Troubleshoot network-layer issues, including VLAN, DNS, DHCP, VPN, NAT, firewall rules, and connectivity failures on host machines

Investigate performance issues involving GPU utilization, container resource constraints, thermal throttling, driver conflicts, and disk I/O bottlenecks

Advise suppliers on installation best practices, including hardware setup, driver configuration, BIOS/firmware settings, and network configuration for optimal performance

Write and maintain internal runbooks, escalation guides, and knowledge base articles to reduce repeat escalations

Build diagnostic and automation tooling in Python and Bash to reduce manual triage overhead

Collaborate with the engineering and support teams to flag and document systemic or recurring platform issues

You Are

Experienced with Linux, especially Ubuntu, and comfortable troubleshooting from the command line

Someone who enjoys debugging difficult problems and fixing broken systems

Methodical and focused on finding root causes, not just temporary fixes

Able to manage complex tickets independently

A clear written communicator with an interest in AI infrastructure and GPU computing

Must-Haves

Strong Linux systems operations experience with Ubuntu, RHEL/CentOS, or Debian, including networking, storage, services, and permissions

Proficiency with Docker, including container debugging, Docker Compose, image management, cgroup limits, and Docker storage and filesystem troubleshooting