Head of Global Compute Capacity & Platform Strategy

Lumalabs Ai

Apply to this job
SF Bay Area, CA hybrid Until 8/22/2026 10+ years exp First posted June 11, 2026 Last posted June 11, 2026
Job description
The Role
Compute is the ultimate physical and financial prerequisite for the robotics foundation models we are building. This role owns Luma’s global compute footprint end-to-end—bridging macro capacity strategy, multi-million dollar capital allocation, and top-tier systems architecture. You will design our scaling roadmap from the silicon up, ensuring our research and robotics teams have the uninterrupted runway they need to ship frontier world models. As a member of the executive team, you will be the single person responsible for turning capital into capability.

What You'll Do
  • Architect Multi-Year Compute Strategy: Lead capacity planning, global vendor and cloud partnerships, on-prem vs. cloud mix, and accelerator supply chain roadmaps (H/B-series GPUs, custom silicon evaluation).
  • Direct the Platform Org: Provide strategic leadership to our infrastructure, distributed systems, and datacenter operations teams—scaling the organization to support next-generation compute demands.
  • Maximize Fleet Utilization: Oversee the architectural efficiency of our cluster configurations to deliver >50% Model Flops Utilization (MFU) on flagship training runs.
  • Command a Megawatt Budget: Negotiate, secure, and operate our largest-scale capital deployments for compute infrastructure, partnering directly with Finance to optimize unit economics and risk management.
  • Unify Global Capacity: Champion the platform strategy that enables world-model training, heavy simulation rollouts, and real-time on-robot inference to seamlessly share a single, elastic fleet.
  • Act as Principal Executive Interface: Serve as the primary commercial and strategic bridge to NVIDIA, AMD, hyperscalers, and frontier silicon vendors.

Qualifications:
  • 10+ years of engineering leadership experience in large-scale distributed systems, infrastructure, or technical supply chain, with a proven track record of leading compute platform strategy at a frontier AI lab, hyperscaler, or major autonomy program.
  • Deep technical & commercial fluency in high-performance cluster topology, high-speed interconnects (InfiniBand/RoCE), large-scale data systems, and the economics of distributed training architectures.
  • Direct operational oversight of 10k+ accelerator environments in high-performance production settings.

Preferred qualifications:
  • Scale Credentials: Experience orchestrating capital or infrastructure for training runs at the >100B-parameter or >100k-GPU-day scale.
  • Robotics/Autonomy Context: Familiarity with the unique capacity and latency demands of edge-to-cloud inference and real-time autonomous systems.

About Luma


Luma’s mission is to build unified general intelligence that can generate, understand, and operate in the physical world.

We believe that multimodality is critical for intelligence. To go beyond language models and build more aware, capable and useful systems, the next step function change will come from vision. So, we are working on training and scaling up multimodal foundation models for systems that can see and understand, show and explain, and eventually interact with our world to effect change.
About this role

Summary

Lead global compute strategy, oversee infrastructure, optimize capacity for AI models

Job title

Head of Global Compute Capacity & Platform Strategy

Experience level

10+ years

Minimum experience

10+ years exp

Industry

technology

Location requirements

SF Bay Area, CA; remote not specified

Salary

Not specified

Management role

Yes

Skills & keywords

Required skills

distributed systemsinfrastructurehigh-performance clusterdata systemssilicon evaluation

Preferred skills

scale infrastructureautonomy systemsedge-to-cloud inference

Specializations

large-scale distributed systemsinfrastructurecompute platformhigh-performance clusterautonomy
Locations

Structured locations inferred from the posting.

Unknown location

Hybrid