The role involves working on frontier foundation models that enhance intelligent experiences at Apple. The team focuses on the full training lifecycle, including pre-training and mid-training approaches tailored for Apple silicon. The work emphasizes solving complex problems in deep learning, particularly in areas like reward modeling and aligning models for various tasks. Candidates will collaborate with a team of engineers and scientists to address challenging issues in natural language processing and voice interactions.

• Demonstrated expertise in deep learning with a publication record in relevant conferences.
• Proficient programming skills in Python and one of the deep learning toolkits such as JAX, PyTorch, or TensorFlow.
• PhD or equivalent practical experience in Computer Science or a related technical field.
• Experience in building end-to-end voice models.
• Experience in speech recognition and generation.
• Experience in reinforcement learning and on-policy distillation.

• Engineer models specifically for Apple silicon.
• Develop mid-training approaches that bridge general capability and task-specific performance.
• Solve frontier problems in reward modeling and handle sparse and delayed rewards.
• Align models reliably across a spectrum of tasks from creative to precise workflows.
• Collaborate with a close-knit team to tackle challenging problems in foundation models and deep learning.

• Employees at Apple are often offered comprehensive benefits that support physical and mental well-being—flexible medical plans, confidential counseling, onsite wellness centers at major campuses, and resources for fitness and daily life. Families typically receive fertility support, paid parental leave with gradual return, caregiving leave, and dependent-care guidance, while financial perks commonly include stock grants (with purchase discounts), 401(k) matching, and income-protection coverage. Employees also see robust time off, Apple University learning and tuition reimbursement, donation matching and paid volunteer hours, and deep product and partner discounts.