Jovan Bubonja

Jovan Bubonja is a software engineer working in Microsoft’s Applied Sciences Group (ASG) Belgrade, where he specializes in model compression techniques, inference optimization, and integration of machine-learning models.  His research interests lie in the field of deep learning, with a focus on model training and model engineering.

Before joining the team, Jovan interned with the Applied Science team within Microsoft’s Office Product Group.  During his internship, he explored encoder-based large-language-model distillation and its applications in multitask learning.  After this, he joined TomTom, a navigation and geolocation company, as an applied scientist, where he contributed to the electric vehicle product team (EV Routing & Range) by researching and enhancing electric vehicle consumption models and developing and maintaining data pipelines.

He holds both a bachelor's and a master's degree from the University of Belgrade, where he focused on machine learning and signal processing.  His academic journey was also complemented by a solid foundation in software engineering.

In his free time, Jovan enjoys playing a variety of sports.  He has a long-standing love for basketball, which he has played for most of his life, and he enjoys playing football (soccer), table tennis, and is a passionate skier.