Applied LLM Research Engineer, Input Experience

As an Applied LLM Research Engineer, you will enable next-generation AI applications using Apple Foundation Models. You will sit at the intersection of cutting-edge research and product reality, bridging the gap between raw model performance and the nuanced needs of Apple customers worldwide. You will explore, design, and implement emerging techniques, ensuring alignment with product goals, privacy requirements, and performance metrics. You will contribute to all phases of model development: problem formulation, experimentation, evaluation, fine-tuning, and continuous improvement. Finally, you will help define and refine new features that expand both the depth of Apple Intelligence’s capabilities and the breadth of its support for our global customer base. Minimum Qualifications PhD in CS/EE/Physics/Statistics/etc.; or Bachelor’s or Master’s in CS/EE/Physics/Statistics/etc combined with 2 years of relevant experience Strong foundations in ML & LLM, including core principles, techniques and practical applications Familiarity with post-training techniques such as SFT, RLHF, data synthesis, Parameter-Efficient Fine-Tuning Familiarity with training frameworks such as PyTorch, JAX, TensorFlow, or equivalent Preferred Qualifications Experience with fine-tuning and deploying large ML models for real world products Experience curating, filtering, and synthesizing high-quality training datasets at scale Experience developing and training models for agentic workflows, tool calling and advanced reasoning techniques Experience with training LLMs with RLVR, reward modeling, environment design Familiarity with training for computer-use capabilities Familiarity with designing hardware-efficient model architectures, optimizing inference latency, and implementing advanced decoding strategies, speculative decoding Active contributor to complex, large-scale codebases, with a strong emphasis on writing high-quality, maintainable, and well-tested code. Experience using AI-assisted development tools (e.g., Claude, Copilot, or similar) to accelerate experimentation, code development, and research workflows Excellent programming and communication skills