I am a Junior at Pace Junior Science College learning Computer Science, Physics, Chemistry and Math. I conduct research on Spatial Perception and Depth Mapping Systems in Computer Vision and work towards enabling quantisation of Machine Learning Models deployed on Embedded Systems.
While my Research is centred around software approaches to Spatial Perception or Computer vision, I work towards enabling these algorithms to equip with Embodied Intelligence. Over time, my research evolved from a classification and detection approach to events in the surrounding to a quantification approach to trying to simulate how these systems would evolve temporally and how such a system could be dployed to machines to give them human-like perception of environment. I’ve experimented with different modalities of haptics, acoustics and orientation apart from computer vision. I like to develop application-oriented research which has a practical use, or contributes in any way to make a positive impact over the next decade. I’ve worked on interdisciplinary projects with Computer Science at the intersection of Plant Phenotyping, Environmental Conservation and also Cognition. I explore how interdisplinary approaches can help create sustainable solutions and ecosystems involving Computer Vision.
I apply my research in the Startup industry by working as a Machine Learning Intern at EdgeImpulse, where I work on enterprise applications to bring Embedded Machine Learning from Remote Agricultural Units to the Industry 5.0. I’ve previously worked as a Summer Intern at Luxonis Corporation where I was introduced to the world of Spatial Perception and 3D applications of computer vision which have become an integral part of my research today. The process or learning, exploring and experimentation eventually led me to explore other domains where Computer Vision could be of practical use. Over this iterative process of learning, I got the opportunity to conduct Research at SERI, Harvard Medical School under Prof. Ayush Kumar where I assist in developing computer vision algorithms for collision estimation for objects gazed upon to study the dwell time and behavior of hemineglect people.