About the role

Anthropic's compute fleet is one of the largest and most varied in the world, and everything we do from training frontier models to serving Claude depends on getting the right work onto the right hardware at the right time. Our Scheduler team owns that problem. We build the scheduling layer for Anthropic's Kubernetes fleet, the tools researchers and engineers use to launch and manage their jobs, and the systems that make sure the fleet is used as efficiently as possible. When a researcher starts a run, the scheduler decides where it lands and how quickly; when demand outstrips supply, it decides who waits. This team provides the paved path that lets everyone at Anthropic get compute when they need it without becoming an expert in the infrastructure underneath.

We're looking for an engineering manager to lead this team. The scheduler is on the critical path for nearly all of Anthropic's compute, and the mandate is expanding quickly: making scheduling work seamlessly across a growing, heterogeneous fleet; raising utilization while keeping jobs starting fast; making the system's decisions predictable and explainable to the people who depend on it; and making the job-launch experience something researchers rarely have to think about. You'll lead a team building infrastructure that the entire research and product organization depends on, and you'll partner closely with capacity planning, research, inference, and product teams to make efficient use of the fleet.

Key responsibilities

• Lead and grow a team of engineers building Anthropic's scheduling platform, job-launch tooling, and fleet-efficiency systems, owning planning, execution, and delivery against key milestones
• Set technical direction for scheduling, placement, queueing, and quota across Anthropic's compute fleet
• Partner with capacity planning, research, inference, and product teams to bring workloads onto the paved path and make efficient scheduling decisions
• Drive the roadmap for scheduler capabilities, fleet utilization, and the developer experience of launching and managing jobs
• Define and track the metrics that measure fleet efficiency and scheduling quality (utilization, queue wait, job-start latency, etc.) and hold the team accountable to them
• Create clarity for the team and stakeholders in an ambiguous, fast-moving environment where demand for compute routinely exceeds supply
• Take an inclusive, equitable approach to hiring, coaching, and career development, and sustain a high-performing, healthy team
• Represent the team across the engineering organization and contribute to engineering-wide initiatives as a member of Anthropic's engineering management group

Minimum qualifications

• Experience managing and growing a team of software engineers
• A hands-on software engineering background as an individual contributor prior to moving into management
• Experience building or operating large-scale distributed or infrastructure systems in production
• Working knowledge of Kubernetes and cluster scheduling concepts, such as resource requests and limits, affinity, priority and preemption, and custom schedulers or controllers
• Excellent written and verbal communication skills, including the ability to create clarity across teams

Preferred qualifications

• 5+ years of engineering management experience, including leading infrastructure, platform, or compute teams
• Experience owning a cluster scheduler, job orchestration system, or resource manager at scale
• Familiarity with scheduling ML workloads on accelerators and the tradeoffs between utilization, fairness, and latency
• Experience building developer tooling that other engineers rely on every day
• A background in observability or incident response for control-plane systems, and a track record of improving production reliability
• A track record of building a culture of belonging and of engineering excellence
• Low ego, high empathy, and a habit of leading by example