San Francisco Bay Area
Homepage: www.isikdogan.com
• Leading machine learning research in generative media enhancement and fidelity. • Working on image and video processing, super resolution, restoration, compression, HDR, and color science, with partners across creative, studio, product, research, and engineering.
• Initiated moonshot projects, proposed novel ideas, ran experiments, and built early prototypes. • Collaborated with design and engineering teams to ready early-stage projects for production. • Built machine learning models for Apple Vision Pro and iPhone.
• Designed neural network architectures for constrained environments, minimizing hardware costs. • Built the machine learning models behind the Intel Eye Contact Corrector, which was featured on numerous media outlets including Fortune Magazine and TechXplore. • Co-developed the VisionISP: an image processing pipeline for computer vision applications.
• Designed and trained models for smart camera features, including landmark recognition on Motorola phones. • Modeled human judgments of image quality and detected several types of image degradation.
Research Advisors: Alan Bovik & Paola Passalacqua • Created DeepWaterMap, a pioneering deep learning model for surface water segmentation. • Developed image processing tools for satellite imagery, including RivaMap and DeepRiver.