Senior Engineering Manager, Compute

Responsibilities

• Strategic direction for Compute: Own the strategy and standards of excellence for the compute layer that the world's agents run on, across design, delivery, and operations. Build a culture of ownership, quality, and customer-first decision-making.
• Tech leadership: Lead, hire, and grow a high-ownership team; roll up sleeves, ready to do deep into the trenches, by staying close to design docs and code, rather than managing from a distance. Coach engineers, level them up, and clear the friction that slows them down.
• Roadmap & trajectory: Drive the arc from today's compute toward the next-generation of compute platforms. Ground prioritization in customer and design-partner feedback, and turn ambiguous, fast-moving requirements into predictable, iterative delivery.
• Operational excellence: When you run frontier AI in production, reliability is the product. Own operations, run on-call and incident response, and drive blameless postmortems and the systemic fixes that prevent recurrence.
• Technical depth: Guide the hard architectural decisions for large-scale, multi-tenant compute, where technical concerns cut across workload isolation and security, scheduling, fleet efficiency / utilization / goodput, and performance, while ensuring the platform is reliable and efficient for the workloads that depend on it.
• Capacity, supply & economics: Own utilization, capacity and supply planning, and the cost-per-unit-of-compute and margin profile of the fleet, across CPU compute today and accelerated compute ahead.
• Cross-team & customer execution: Partner with leadership, Product, SDK, UX/DX, Security, and design-partner customers to align priorities and unblock delivery. Communicate progress, tradeoffs, and risk clearly to technical and non-technical audiences alike.

Qualifications

• Proven experience leading software engineering teams that build and operate large-scale compute platforms or fleets, with strong operational practices.
• 12+ years in software and/or infrastructure engineering, including 7+ years of people management and demonstrated ownership of delivery and live-site outcomes.
• Deep distributed-systems and compute infrastructure depth, with the hands-on judgment to guide architecture and execution rather than from a distance.
• Experience operating multi-tenant compute that other people's production workloads depend on.
• Bachelor's degree in Computer Science or related field, or equivalent practical experience; advanced degree a plus.
• Excellent communication skills, with the ability to partner across engineering, product, and leadership and fold customer feedback into the roadmap.

Required Skills

• Strong leadership, coaching, and performance management; ability to grow engineers and build a healthy, accountable, high-ownership team.
• Excellence in execution: planning, prioritization, and delivering iterative milestones in an ambiguous, fast-moving environment while managing unplanned work.
• Fleet thinking: utilization, goodput, capacity and supply planning, and cost discipline as first-class engineering concerns.
• Live-site reliability craft: on-call, incident management & response, and postmortem-driven continuous improvement.
• Strong command of the building blocks of a compute platform: multi-tenant isolation and security, scheduling, and resource management.
• Ability to review and raise the bar on technical artifacts (design docs, code reviews) across a distributed-systems codebase.

Preferred Experience

• MicroVMs and virtualization (Firecracker, gVisor, Edera) or managed-compute primitives (AWS Fargate, GCP Cloud Run, AWS Lambda), and/or Kubernetes internals.
• Building serverless or hosted-compute products from 0 to 1, including the rapid-delivery-vs-durable-platform tradeoffs that come with it.
• Multi-cloud delivery across AWS and GCP.
• Cold-start, warm-pool, and scheduling/latency optimization for on-demand compute.
• Agent sandboxes, secure execution of untrusted code, or other AI-agent infrastructure.
• GPU / accelerated compute: fractional GPUs (MIG, MPS, time-slicing), GPU scheduling, training vs. inference fleets, and multi-tenant GPU isolation.