Machine Learning Engineer, Computer Vision

Apple • Cupertino, CA

7h

About The Position

The Machine Learning Engineer will build computer vision models and vision-language models (VLMs) that enable Apple devices to recognize and understand objects around the user. The team develops foundational perception technologies that are reused across experiences and devices, including iPhone, iPad, Mac, and HomeView. These technologies power Apple Intelligence features such as Visual Intelligence and Clean Up in Photos. The work spans lightweight, task-specific vision models and general-purpose VLMs, translating recent research into models deployed on millions of devices.

The role focuses on designing, training, and improving models for object understanding and recognition; ensuring they perform robustly on real-world data and across devices; defining product-quality evaluations; and accounting for on-device memory, latency, and power constraints.

Requirements

• Ph.D. or Master's degree in Computer Science, Machine Learning, Computer Vision, or a related field, or equivalent industry experience.
• Strong foundations in computer vision and deep learning.
• A track record of making models work in practice, including end-to-end projects covering data, training, and evaluation on real-world visual data.
• Proficiency in Python and PyTorch or an equivalent deep learning framework.
• Proficiency in using generative coding tools.
• Rigorous experimentation skills, including clear baselines, ablations, and meaningful evaluation sets.

Nice To Haves

• Publications in top computer vision or machine learning venues such as CVPR, ICCV, ECCV, NeurIPS, or ICLR, ideally with released code or work transferred to a product.
• Experience in representation learning, such as self-supervised learning, metric learning, retrieval, or recognition.
• Experience applying machine learning in an industry or product setting.
• Experience with knowledge distillation or model compression.
• Familiarity with vision-language models.

Responsibilities

• Design, train, and improve computer vision models and VLMs for object understanding and recognition.
• Make models robust to real-world conditions and across devices, and address failure modes found in real data.
• Define evaluations that reflect product quality and use them to guide model decisions.
• Design models for on-device memory, latency, and power constraints.
• Translate recent research into models that ship on millions of devices.

What This Job Offers

• Job Type: Full-time