Sr. Applied Scientist, AI Evaluation & Quality Systems

Apple Services Engineering (ASE) powers the AI and LLM features behind experiences that hundreds of millions of users love every day. As these systems increasingly rely on human-in-the-loop evaluation, the quality of our products is directly constrained by the quality of our evaluation systems. We believe that to build exceptional AI, you need exceptional mechanisms to validate the signals used to train and evaluate them. The Human-centered AI, ML Data Quality Operations team is looking for a Senior Applied Scientist to join our growing team. We are building the systems and methodologies that make AI evaluation trustworthy, and scalable - directly shaping how Apple develops and validates AI across products and services. In this role, you will develop novel, scalable quality control solutions, working closely with cross-functional teams to ensure the data powering our AI/ML systems meets the highest standards of accuracy, consistency, and relevance. Your work will span the full lifecycle of quality assurance for AI and human judgments - from real-time validation and human-verified ground truth generation, to root-cause analysis that turns disagreements into corrective action. This role demands fluency across research thinking and engineering execution - you will prototype, validate, and ship. A strong point of view on when not to use a model or agent is as valued here as the ability to build one. Design and implement scalable ground truth generation pipelines across varied task types, annotation modalities, and cold start conditions Build and maintain real-time monitoring systems that detect drift, distribution shifts, and quality degradation as they emerge across live evaluation and annotation pipelines. Design calibration frameworks that periodically re-anchor LLM evaluators against human-verified gold sets, correcting drift before it compounds. Build root-cause analysis tooling that surfaces disagreement patterns between automated and human judgments, and feeds findings directly into annotator training and guideline refinement. Partner closely with downstream users of these systems -ML teams, LLM-as-a-Judge (evaluator) developers, annotators - to ground design decisions in real feedback and usage patterns, not just architecture. Communicate findings and recommendations clearly to both technical and non-technical stakeholders5+ years of industry experience in applied science or machine learning, with demonstrated experience building or operating production-grade evaluation, annotation, or quality-assurance pipelines. Hands-on experience designing ground truth generation pipelines across varied task types and annotation modalities, including cold-start scenarios with limited existing data. Experience building real-time monitoring or anomaly/drift detection systems for live data or ML pipelines. Working knowledge of evaluation methodology for generative AI - including LLM-as-a-judge design, meta-evaluation, failure mode analysis, and calibration/reference-guided grading techniques Strong software engineering fundamentals and proficiency in Python and relevant ML frameworks, with production experience building, deploying, and monitoring LLM-based pipelines and agents. Demonstrated ability to work directly with downstream users/stakeholders to incorporate feedback into system design, and to communicate findings clearly to both technical and non-technical audiences. MS or PhD in Computer Science, Machine Learning, Statistics, or a related quantitative field, or equivalent practical experience.PhD in Computer Science, Machine Learning, Statistics, or a related field Experience in designing systems or tooling that are configurable and extensible by practitioners who did not build them Strong communication skills with the ability to influence technical direction across cross-functional teams Demonstrated passion for leveraging AI to improve work efficiency and scale