Senior Applied ML Scientist – Generative Video

Apple • Cupertino, CA

2d

About The Position

We are looking for a passionate Senior Applied ML Researcher to drive innovation
in generative video modeling, with a strong focus on diffusion-based methods and
related generative features. In this role, you will research, design, and deploy
models at the intersection of machine learning research and real-world product
impact. You will collaborate closely with research scientists, engineers, and
product teams to find novel applications of generative AI capabilities to assist
our creative user base. Your mission is to elevate the workflows of millions of
creators by combining generative AI with Apple’s human-centered design
principles.

DESCRIPTION

Design and train state-of-the-art generative video models, primarily based on
diffusion, consistency, rectified flow, or related generative frameworks Explore
novel architectures for spatiotemporal modeling (e.g., 3D U-Nets, DiT-style
Transformers, hybrid CNN-Transformer models) Conduct experiments on long-range
temporal coherence, motion consistency, controllability, and multi-modal
conditioning (text, audio, images)

Requirements

• MS in Computer Science, Machine Learning, or a related field, or equivalent practical experience
• 4+ years of experience in deep learning for generative models, particularly diffusion-based methods
• Hands-on experience in distributed training of large models using PyTorch (or equivalent frameworks)
• Solid understanding of video representations, spatiotemporal modeling, and neural network optimization.
• Experience designing & training multi-modal DiT-style Transformers, latent diffusion, or multi-stage video generation pipelines
• Experience design & training adapters
• Proven ability as a creative problem solver

Nice To Haves

• PhD with research focused on generative modeling, diffusion models, or video understanding
• Experience with text-to-video, image-to-video, or video-to-video generation
• Publications in top-tier ML conferences (e.g., NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, etc.)

Responsibilities

• Design and train state-of-the-art generative video models, primarily based on diffusion, consistency, rectified flow, or related generative frameworks
• Explore novel architectures for spatiotemporal modeling (e.g., 3D U-Nets, DiT-style Transformers, hybrid CNN-Transformer models)
• Conduct experiments on long-range temporal coherence, motion consistency, controllability, and multi-modal conditioning (text, audio, images)

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Number of Employees: 5,001-10,000 employees