Michael James Kastoryano serves as an Associate Professor in the Department of Computer Science at the University of Copenhagen, specializing in Machine Learning research within the department's dedicated Machine Learning section. His work addresses critical challenges in model efficiency and scalability for modern AI systems. His core research spans Machine Learning, Deep Learning, and Computer Vision with emphasis on creating compact representations through tensor decomposition and quantization techniques. Current investigations focus on optimizing visual recognition pipelines and language model pretraining under resource constraints. Analysis of his 2024 publications reveals a consistent trajectory toward efficient deep learning architectures, particularly in visual representation learning and quantized adaptation methods. These contributions target practical deployment barriers for AI models on edge devices while maintaining performance integrity across computer vision and NLP domains.
