3+ years of experience building production ML systems or ML infrastructure Strong programming skills in Python and/or Rust/Java Understanding of end-to-end machine learning workflows - from data preparation through training, evaluation, and deployment Experience with distributed systems and large-scale data processing Experience with model serving, inference optimization, or ML pipeline engineering Experience building APIs and services that other engineers consume Strong collaboration and communication skills Comfortable navigating ambiguity in fast-moving areas BS, MS, or PhD in Computer Science or equivalent practical experienceExperience with LLM inference optimization (batching, quantization, KV caching, tensor parallelism) Experience with model serving frameworks (vLLM, TensorRT, Ray Serve, or similar) Experience with embedding models and retrieval systems - fine-tuning encoders on graded or contrastive objectives, pooling strategies, dimensionality reduction for serving cost, vector databases, and retrieval evaluation (NDCG, recall, graded relevance) Experience with fine-tuning and alignment workflows (SFT, DPO, LoRA, RLHF, RLVR, GRPO, reward modeling) Experience with feature engineering and feature serving platforms (e.g. As a member of the team, your responsibilities will include: Design, build, and optimize large-scale ML platform services used by teams across Apple Build the embedding and retrieval path end to end - fine-tuning encoder models, encoding corpora at scale, building and serving vector indexes, and evaluating retrieval quality so improvements are measurable rather than asserted Build and operate the feature store teams use for training and serving, keeping both paths consistent off a single feature definition Develop optimization capabilities that reduce cost and improve quality across ML workloads - including model routing, caching, serving configuration, inference optimization, and training efficiency Build managed, self-service experiences so customers can go from data to production AI with minimal friction Build managed training - supervised fine-tuning, reinforcement learning and distillation - so teams can customize models without running their own training infrastructure Build governance and compliance capabilities - lineage, policy enforcement, cost observability, and access control Partner with customer teams across Apple to understand their ML workloads and deliver production solutions Operate production services with on-call responsibilities.