Computer Vision Researcher - Generative Models for Video Restoration
Apple
1 day ago
On-site
Herzliya, Tel Aviv District, United States
We are seeking a talented Computer Vision / ML Researcher to develop diffusion-based generative models for video restoration in new imaging systems. The role will focus on large, high-quality models that restore degraded video while preserving fidelity, detail, motion, and temporal consistency. You will be part of an interdisciplinary team developing technologies that will shape future Apple products.\\nDeep hands-on experience with diffusion models is required. The ideal candidate has worked on diffusion-based image or video models and has a strong research track record, ideally including publications on diffusion, generative vision, video models, restoration, or inverse problems.
Deep learning development: Design, train, evaluate, and optimize deep learning models for video restoration tasks such as denoising, deblurring, artifact removal, super-resolution, detail recovery, and temporal stabilization. Work with large-scale video datasets, synthetic and real degradations, and rigorous evaluation pipelines.\\nResearch: Develop and adapt diffusion models for video restoration, including video diffusion, DiT-style architectures, latent diffusion, conditional diffusion, diffusion for inverse problems, and efficient sampling. Analyze results, compare against leading methods, and investigate approaches for fidelity preservation, hallucination control, temporal consistency, distillation, and production-quality inference.\\nCollaboration: Work closely with ML researchers, engineers, and cross-functional teams to translate research ideas into robust models and practical systems.
Strong foundation in computer vision, machine learning, deep learning, and video processing.\\nDeep hands-on experience with diffusion models.\\nExperience training diffusion-based image or video models.\\nProficiency in Python and deep learning frameworks such as PyTorch.\\nHands-on experience training deep learning models using large-scale datasets.\\nExperience with model evaluation, debugging, experimental analysis, and failure analysis.\\nMaster’s or PhD in Computer Science, Electrical Engineering, Machine Learning, Computer Vision, or a related field, or equivalent experience.\\nStrong written and verbal communication skills.\\nAbility to work both independently and collaboratively.\\n5+ years of relevant experience, or a PhD with relevant research contributions.
Publications on diffusion models, generative vision, video models, restoration, or inverse problems at top-tier venues such as CVPR, ICCV, ECCV, NeurIPS.\\nExperience with video diffusion, DiT architectures, latent diffusion, conditional diffusion, rectified flow, consistency models, or diffusion distillation.\\nExperience with video restoration, super-resolution, denoising, deblurring, artifact removal, inverse problems, or computational imaging.\\nStrong understanding of temporal consistency, motion, occlusion, flicker, hallucination control, and fidelity-preserving generation.\\nExperience with efficient inference, model optimization, distillation, or deployment on constrained hardware.\\nBackground in signal processing, physics, computational imaging, or inverse problems.
Deep learning development: Design, train, evaluate, and optimize deep learning models for video restoration tasks such as denoising, deblurring, artifact removal, super-resolution, detail recovery, and temporal stabilization. Work with large-scale video datasets, synthetic and real degradations, and rigorous evaluation pipelines.\\nResearch: Develop and adapt diffusion models for video restoration, including video diffusion, DiT-style architectures, latent diffusion, conditional diffusion, diffusion for inverse problems, and efficient sampling. Analyze results, compare against leading methods, and investigate approaches for fidelity preservation, hallucination control, temporal consistency, distillation, and production-quality inference.\\nCollaboration: Work closely with ML researchers, engineers, and cross-functional teams to translate research ideas into robust models and practical systems.
Strong foundation in computer vision, machine learning, deep learning, and video processing.\\nDeep hands-on experience with diffusion models.\\nExperience training diffusion-based image or video models.\\nProficiency in Python and deep learning frameworks such as PyTorch.\\nHands-on experience training deep learning models using large-scale datasets.\\nExperience with model evaluation, debugging, experimental analysis, and failure analysis.\\nMaster’s or PhD in Computer Science, Electrical Engineering, Machine Learning, Computer Vision, or a related field, or equivalent experience.\\nStrong written and verbal communication skills.\\nAbility to work both independently and collaboratively.\\n5+ years of relevant experience, or a PhD with relevant research contributions.
Publications on diffusion models, generative vision, video models, restoration, or inverse problems at top-tier venues such as CVPR, ICCV, ECCV, NeurIPS.\\nExperience with video diffusion, DiT architectures, latent diffusion, conditional diffusion, rectified flow, consistency models, or diffusion distillation.\\nExperience with video restoration, super-resolution, denoising, deblurring, artifact removal, inverse problems, or computational imaging.\\nStrong understanding of temporal consistency, motion, occlusion, flicker, hallucination control, and fidelity-preserving generation.\\nExperience with efficient inference, model optimization, distillation, or deployment on constrained hardware.\\nBackground in signal processing, physics, computational imaging, or inverse problems.