Braden Refalo
Braden Refalo
Braden Refalo is a PhD candidate at TU Delft in the Embedded Systems group, working on embedded AI. As part of the ANT project, his research explores binarising machine learning models, such as large language models, to improve their efficiency, alongside kernel-level optimisations. His work aims to enable advanced models to run effectively on low-end hardware and in distributed settings, where models can be split across multiple devices or cloud systems to aggregate compute resources through split inference.
Braden holds an MSc in Artificial Intelligence with distinction and a BSc (Hons.) in Information Technology in Artificial Intelligence with first class honours from the University of Malta. During his studies he focused on multimodal models and ensemble learning techniques, primarily working with transformer-based architectures.
He is passionate about AI systems that are efficient, private and explainable, which he sees as key principles for the development of responsible and accessible AI. He advocates for a future where AI technologies are available to everyone and is particularly motivated by the societal impact AI can have across domains such as biomedical applications, rehabilitation and beyond.