Staff AI Infrastructure Engineer

About Luma AI

A new class of intelligence is emerging, systems that understand and generate the world across video, images, audio, and language. Building multimodal AGI is not just a modeling challenge. It is an infrastructure challenge at the edge of what hardware, software, and organizations can support. At Luma, we operate rapidly scaling 10k+ GPU fleets, pushing utilization, throughput, and reliability hard enough that yesterday’s solutions break regularly. Researchers depend on this infrastructure to move the frontier forward. Customers depend on it to power real creative work. Many companies run accelerators. Very few sit directly next to the teams inventing the models that redefine what those accelerators must do. At Luma, improvements to scheduling, efficiency, and reliability immediately translate into faster research iteration and entirely new product capabilities. We are still early. The playbook is still being written.

What You’ll Own

• Reliability of the Frontier
• Architect and operate large, heterogeneous GPU environments under extreme demand
• Improve utilization and performance where small gains materially change company outcomes
• Resolve failures that span hardware, OS, runtimes, and orchestration
• Eliminate entire classes of instability
• Build mechanisms that make heroics unnecessary
• Scheduling, placement, and resource management for increasingly complex jobs
• Design scheduling, placement, and resource management approaches for increasingly complex jobs
• Work directly with research to build the systems required for new model capabilities
• Ensure inference platforms scale rapidly without sacrificing reliability or latency
• Anticipate where today’s abstractions will fail and redesign ahead of them

Building the Organization

• Hire and develop exceptional systems and reliability engineers
• Set the bar for technical depth, judgment, and production ownership
• Shape architecture early through strong partnerships with research and product
• Translate reliability constraints into long-term platform strategy

Required

• Deep expertise in Linux and distributed systems
• Experience operating GPU / accelerator clusters in real production environments
• Strong fluency in Kubernetes and modern open-source infrastructure
• Comfortable debugging across hardware → kernel → runtime → orchestration
• You understand how systems behave under contention and at scale
• You write code and build automation
• You think in bottlenecks, failure modes, and tradeoffs
• Engineers trust your judgment, especially when things break

Leadership Expectations

• Raise reliability standards across the company
• Influence product and research architecture early
• Build strong partnerships, not ticket queues
• Attract and level up exceptional engineers
• Be curious how models use infrastructure, because improving systems expands what becomes possible

Why This Role Is Special

• This role helps define how reliability works for a new generation of AI infrastructure
• The decisions you make here will influence how research progresses
• How products scale
• How customers trust us
• And how the engineering organization grows

Important

This role requires comfort operating close to upstream and close to the metal. If most of your experience has been inside highly abstracted internal platforms where others owned the underlying machinery, this is unlikely to be a match.