• Can work independently with minimal hand-holding

Support existing product / speed up development

The company builds and operates large-scale GPU infrastructure to power AI workloads for its customers. The team is building agentic automation so it can run a GPU fleet 10x its current size without growing headcount at the same pace.

Internally, this role is called AI Infrastructure Engineer: the team sees its SREs as strong software engineers and architects, not traditional ops. You’ll keep production systems and the GPU fleet healthy while building the automation that makes that work scale. Most of the role is engineering, with smaller part being infrastructure operations (node diagnostics, hardware verification, cluster operations). There’s no customer-facing support.

You’ll join the full-time infrastructure team as a contractor. Strong performers will take on broader automation and design work, and top performers will be considered for full-time positions.

Requirements

• 5+ years in infrastructure, SRE, or systems engineering
• Recent, hands-on software engineering experience: you build and maintain real software, not only scripts and configs
• Strong Python or Go
• Kubernetes — hands-on cluster and workload management
• Strong Linux systems administration
• Configuration management experience — Ansible preferred; Puppet or Chef acceptable
• Automation scripting to eliminate repetitive tasks
• Ability to follow and execute detailed operational runbooks
• Proficiency in programming/scripting languages
• Direct experience in monitoring and observability practices
• Advanced knowledge of cloud services
• Strong AI usage skills, but also ability to solve problems without AI tools when needed
• Bare-metal data center operations
• Ability to thrive in a collaborative environment involving different stakeholders and subject matter experts

Nice to Have

• InfiniBand or high-speed networking (RoCE / RDMA)
• Terraform and cloud infrastructure knowledge
• GPU/NVIDIA GPU experience — firmware, hardware faults, diagnostics

Responsibilities

• Design and build automation, including agentic tooling, to replace manual operational work across the GPU fleet
• Diagnose and resolve infrastructure issues on compute nodes and clusters (internal tickets, not customer support)
• Run and automate node lifecycle operations: prepare, test and return nodes from repair (RMA) to production
• Gather logs and diagnostic data for hardware vendors during RMA
• Support new cluster bring-ups as the fleet grows
• Work with data center teams on hardware verification and troubleshooting

How Success is Measured

• Number of support tickets resolved per week
• Average time-to-resolve per ticket
• Nodes returned to active fleet per sprint
• Automation improvements shipped

Important:

• The "tickets" refer to internal infrastructure and operational work rather than customer support. You'll mainly be diagnosing issues with compute nodes and infrastructure that are linked to support tickets. You may collaborate with data center teams on hardware verification and troubleshooting, but you won't be interacting directly with customers.
• The role is primarily engineering-focused. A good estimate is: 80–90% engineering, 10–20% operational/support work. Even the support work is mostly infrastructure-related (node diagnostics, hardware verification, cluster operations), not customer-facing support.

• 1st — technical/experience fit
• 2nd — live coding
• 3rd — troubleshooting + incident response, with a cultural-fit aspect

Slack, video calls, direct collaboration with principal engineers and team leads

10+