Bishnu Dev

MBZUAI, Masdar City

Abu Dhabi, UAE

Nationality: Nepalese

I am Bishnu Dev, an MSc student in Machine Learning at the MBZUAI, advised by Prof. Samuel Horváth. My research spans both theory and practice: on the theory side, I focus on optimization and efficient machine learning; on the practical side, I build sensing and interaction technologies.

Previously, I was a Visiting Scholar at the SPICE Lab at Northwestern University, where I worked with Prof. Karan Ahuja on calibration-free methods for human motion sensing using wearable devices. I earned my BSc with a double major in Computer Science and Mathematics from New York University Abu Dhabi, where I worked with Prof. Christina Pöpper on privacy-preserving web search and with Prof. Nizar Habash on interactive video systems for multilingual communication.

Publications / Research Collaboration

2023

Boundless Conversations: AI-Powered Video Interactions across Domains, Languages, and Time

Alberto Chierici, Soojin Lee, Nizar Habash, and 4 more authors

In SIGGRAPH Asia 2023 Emerging Technologies, 2023

Abs

We present novel web applications that blend AI and video communication, offering an interactive way to engage with videos, transcending time and distance constraints. Known as Time Offset Interaction Application (TOIA), the platform transforms passive video viewing into dynamic conversations using personal recordings or YouTube clips. Our work expands TOIA with two unique experiences: 1) "The Elephant in the Room" project, stimulating dialogues on sensitive topics like childbirth, sex, and death; 2) Support for multilingual interactions, fostering a global collection of diverse personal narratives. Through AI and interactive media, we aim to broaden the horizons for sharing human experiences and perspectives.