The AWS Data Center Availability Serviceability team works with hundreds of AWS data centers globally to deliver the highest quality and lowest cost availability, capacity, and scaling results for our customers. We standardize operations globally by delivering tools, policy, processes, and procedures to our internal teams.
Serviceability is the ability to maintain, repair, and manage infrastructure assets efficiently throughout their lifecycle -- encompassing maintenance standards, spares management, training and certifications, operational readiness, and the systematic use of data and automation to prevent disruptions and drive continuous improvement.
We are seeking a Maintenance Optimization Technical Program Manager to own how AWS data center infrastructure is maintained: which tasks, at what interval, with what controls, and why. The fleet runs thousands of preventive maintenance plans inherited from vendor manuals and historical practice. This role applies reliability-centered maintenance thinking to that inventory at fleet scale, deciding what each task protects against, whether the control fits the failure mode and its consequence, and where a human check can be replaced by telemetry or retired altogether.
Key job responsibilities
You will lead preventive maintenance optimization across equipment classes (electrical and mechanical), taking each change from failure-mode analysis through a revised maintenance standard to the published procedures technicians execute, with the labor and risk impact quantified. You will set the method: a repeatable, lightweight way to challenge a maintenance task (failure mode, consequence, existing detection, proposed control) that program managers across the team can run without you. Working with our monitoring, controls, and maintenance-planning teams, you will drive decisions on where remote telemetry can replace a physical inspection, removing a human check only when the data demonstrably covers it.
You will own decisions that most maintenance organizations never revisit. The problem is real engineering, not process administration: you will be asked why a task exists, what happens if it stops, and whether the data can prove it. You will work across EAM data, equipment telemetry, regulatory regimes, vendor warranty terms, and the physical gear itself, and you will do it inside the constraints of a live production fleet where being wrong has consequences. And you will build the capability in others. The TPMs you coach will carry this reasoning into every program they touch.
