Devansh Bisla

PhD Candidate

Connect

Devansh Bisla

Research Interests: Machine Learning, Deep Learning, and Computer VIsion

Tandon School of Engineering, New York University - 2018

Master of Science, Electrical Engineering

Manipal Institute of Technology, Manipal University - 2016

B.Tech, Electronics and Communications


Hearst - New York

From: June - Aug, 2017

Machine Learning Intern

Incorporated fast content-based image retrieval into Hearst's digital asset management system consisting of over 13M images. We utilized a product quantizer based nearest neighbor search on image features extracted from models pre-trained on ImageNet.

 

NVIDIA - New Jersey

From: May - Aug, 2017

Deep Learning Software Intern

Training on large datasets certainly improves the performance of the network but such datasets often contain a lot of redundant samples. This redundancy leads to a significant increase in training time and computational requirements with minimal gains in performance. In order to remove redundant information, we implemented a sampling methodology hinged upon the geographical distance between two training samples. The implemented methodology resulted in a 50% reduction in training dataset size (50% increase in training speed) with a <5% reduction in peak performance.


Dept of Electrical Engineering, NYU

From: Jan - May, 2020

Teaching Assitant: Introduction to Machine Learning

Center of K12 Stem Education, NYU

From: May - Aug, 2019

Mentor 

Undergraduate Summer Research Program, NYU

From: May - Aug, 2019

Mentor

Center of Urban Science and Planning, NYU

From: May - Dec, 2017

Teaching Assitant: Urban Decision Models, Operations Research for Cities and Leading with IT.


School of Engineering Fellowship - 2018

Graduate Student Scholarship - 2016