Fleet Management Engineer

Fleet Management builds the automation layer for Meta's data center fleet. We own the systems that plan, orchestrate, and execute the operational workflows that keep hundreds of thousands of machines healthy, available, and efficient across every region, and increasingly across public cloud providers as well. Our platforms cover the full lifecycle of work performed on the fleet:

• Planned maintenance and rollouts: firmware, kernel, OS, and hardware maintenance executed safely and predictably at fleet scale.
• Unplanned work and repair: detecting failures, deciding what to do about them, and driving the repair workflow through until capacity returns to production.
• Fleet lifecycle operations: turn-ups, decommissions, server and service moves, rack logistics, and capacity rebalancing.
• Safety and capacity control: deciding how much of the fleet may be unavailable at once, resolving conflicts between competing operations, rate limiting, graceful cancellation, and emergency stop.
• Autonomy: replacing human-driven operational decisions with automation and agentic workflows, backed by the observability and quality signals needed to trust them.

Practically, this means we build large-scale workflow orchestration engines: systems that model operational intent, schedule it against fleet constraints, execute it across dozens of downstream systems, and decide what to do when steps fail. The engineering problems are distributed systems problems - state machines, scheduling, conflict resolution, idempotency, partial failure, and strong safety guarantees over irreversible physical actions. The fleet is growing fast, and the operational volume with it. We are not going to staff our way through that, so our mandate is to make fleet operations scale by an order of magnitude without a corresponding increase in human effort.

Responsibilities

• Own technical direction for a significant problem area of fleet automation – define the multi-year architecture, decide what to build, consolidate or deprecate, and drive it through to production
• Design and evolve orchestration engines: workflow modelling, scheduling and conflict resolution, execution state machines, and the contracts between our platform and the many systems that actually perform the work
• Make large-scale automation safe. Design the guardrails – unavailability budgets, rate limiting grounded in real fleet health, graceful cancellation, handbrakes, blast-radius control – for operations that physically change production capacity
• Lead through others. Decompose a broad ambiguous area into workstreams other engineers can own, review their designs, and be the technical owner the technical owner that the team's most consequential decisions route through route through
• Drive alignment across a wide cross-functional surface – capacity management, provisioning, remediation, hardware and data center operations, and external infrastructure providers. Much of the work is negotiating and formalising contracts and SLAs between systems and organisations
• Change the operating trajectory. Replace escalation paths, one-off scripts and per-case integrations with platforms and abstractions; make structural reductions in operational load rather than incremental patches
• Raise the engineering bar: mentor other engineers, set standards, and bring credible technical framing into roadmap and investment discussions