ML Compute Efficiency Automation Engineer, Infrastructure & Planning

Apples Platform Acceleration u0026 Compute Efficiency (PACE) is a high-leverage team operating at the intersection of our ML organizations underlying compute infrastructure and core platform tooling. Our mission is to empower Apples software engineering teams with efficient scalable compute. By driving out operational friction and optimizing the broader machine learning ecosystem we directly accelerate the pace of development for our Software and AIML models are central to Apples user experiences and maximizing the efficiency of our ML compute is paramount. Compute efficiency sits at the center of this role ensuring that Apples models run as fast reliably and cost-effectively as this role you will tackle optimization challenges from maximizing hardware utilization across GPUs TPUs and custom Apple Silicon to shaping workload scheduling and capacity allocation for large model are looking for a particular kind of builder an exceptional engineer who can think through hard problems and code their way past them especially the ones involving scale and the slow manual work that quietly drains a high-leverage team. The ideal candidate treats every repeated process as a system waiting to be automated every manual escalation as a system not yet built and every prioritization request as a problem the right tooling can solve faster. The resulting data forms a foundation for the rapid high-quality decisions that empower Apples technical and business is a founding role. The majority of your time goes to AI automation building the systems that turn manual operations into tooling that runs and corrects itself. Your remaining time will go towards hands-on ML compute efficiency working along side senior ML efficiency engineers directly on the optimization problems behind the numbers. You will share ownership of PACEs governance and operations with our tools team who is actively building solutions with AI. The work a traditional operations team would grind through by hand things like resource requests allocation tracking escalations and efficiency reporting you will turn into systems that run themselves and watch themselves. When you have done it well the busywork is gone and PACE moves faster than its size says it should. Your challenging work will result in high development velocity and efficient compute accelerating not only Apple but also your career as well.

- Govern compute as code. Build the systems of record for resource requests allocations and utilization accurate and at scale so leadership can trust the numbers.n- Hunt down ML inefficiency. Dig into inference and training workloads across GPUs TPUs and custom Apple Silicon find where compute is wasted trace it to a cause and drive the fix.n- Work the real optimization problems: scheduling capacity allocation and serving cost alongside the engineers who own those systems.n- Get rid of the toil. Replace the time-sink workflows triage reporting reconciliation with systems that handle the routine and pull a person in only when judgment matters. Drive manual escalations toward zero instead of standing up a tiered on call org.n- Make the data useful. Build the telemetry schemas and anomaly detection that surface efficiency and cost opportunities then wire them into tooling that acts rather than just files a report.n- Rebuild what breaks at scale. When a process buckles under Apple scale ML demand re-architect it so it grows with usage instead of headcount.n- Make a lasting impact. Turn what you build into reusable tooling so the rest of the team benefits without coming back to you each time.

BS in Computer Science Computer Engineering or equivalent practical experience.n6 or more years building production software automation tooling or data and infrastructure problem solver who builds first. You have designed things from scratch to wipe out manual work or get past a scale ceiling and you can show us something you with AI tooling. Coding assistants as part of how you already work not something you read programming skills Python or similar or automation pipelines and plus dashboards or data products in something like Tableau Looker or designing data models or telemetry schemas for infrastructure capacity or utilization running complex systems in a large scale compute cloud or infrastructure knowing where not to automate and how to guardrail systems that act on their cross-team collaborator who moves work forward through influence rather than authority and is comfortable owning systems others rely on daily.

Production experience shipping automated or autonomous workflowsnUnderstanding of ML training and inference infrastructure GPU and TPU utilization training throughput scheduling efficiency and foundation model servingnExperience building automated alerting or anomaly detection for infrastructure metricsnExperience with FinOps capacity planning cloud cost management or IT governancenKnowledge of Django/PostgresnLove for open-ended go figure it out and build it projects