CareerPlan

Head of Compute and Capacity - Lithos Ai

San Francisco Bay AreaFull-time2026-11-04 → 2026-11-19

Core

Own the multi-year capacity plan, procurement, and commercial terms for GPU, network, and data-center resources to support agentic inference at scale.

Role type

Head of Compute and Capacity (Infrastructure)

Builds

GPU clusters, network fabric, and data-center capacity for AI inference workloads

Domain

AI Infrastructure / Cloud Computing / Data Center Operations

Deliverable

infrastructure

Required skills

multi-year capacity planning, vendor procurement and negotiation, contract management (commit profiles, SLAs, service credits), TCO modeling, cost-per-token analysis, high-speed networking knowledge, GPU generation expertise

Preferred skills

AI inference or training workload capacity experience, vendor-side GPU sales experience

Technologies

InfiniBand, RoCE, GPUs (various generations)

Responsibilities

Translate demand forecasts and model roadmap into GPU/network requirements; source and negotiate GPU capacity and colocation across neoclouds and hyperscalers; lead end-to-end contract negotiations including pricing and SLAs; build rolling price books and vendor scorecards; model total cost of ownership per unit of performance; partner with engineering teams on hardware acceptance testing

Seniority

Executive (C-level/VP), hands-on IC