About Me

View my publications here

I am a fifth year Ph.D. student at Rutgers Univeristy in the ECE department, where I focus on the intersection of computer vision, machine learning, robotics, and their impact on society. I'm currently an NSF Fellow on the Reality Aware Networks project. I used to be an NSF Fellow on the SOCRATES NRT. My advisor is Professor Kristin Dana. Before this, I got my bachelor's degree in Electrical and Computer Engineering also from Rutgers University.

Awards

In May 2024, I received the Chancellor's Leadership Award for my involvement on campus. In May 2019, my team and I won third place in the Rutgers ECE Department Capstone Design Showcase.

Research

My current research deals with socially cognizant robotic navigation and latent topological maps. My most recent work, Feudal Networks for Visual Navigation, creates a feudal network for visual navigation with a three level hierarchy that is evaluated on the image goal task. The high level manager acts as a memory module and creates a memory proxy map (MPM) to keep track of historical agent locations. The mid level manager (WayNet) mimics human navigation policies by predicting waypoints to guide agent exploration. The low level worker uses this waypoint to execute low level actions in the environment. We train this network with our Human Navigation Dataset, a new dataset that collects observation and human point-click pairs as a human operator explores simulated environments using Habitat AI.

My previous work pertained to social behavior characterization and human-robot interaction under the SOCRATES NSF NRT. We created a series of networks that learned a social behavior dictionary from aerial images of pedestrians. With this dictionary, we analyzed space usage and social behavior patterns,as well as simplifying the pedestrian trajectory prediction problem in order to use simple MLPs to achieve near SOTA results.

In my work with autonomous vehicle steering angle prediction, we created a network, Feudal Steering, that uses hierarchical networks to exploit the temporal abstraction inherent in driving tasks to make predicting steering angles from dash-cam images easier.

I've also done work on vision‑based, real‑time cranberry albedo analysis for crop ripening predictions, overheating risk analysis, and high throughput phenotyping.

Internships

I've had the opportunity to work on several different projects with other teams. At Apple in 2023, I led a cross‑team effort that incorporated multiple device systems into the photos app for memories and researched factors contributing to a users' connection to their photo library.

At Apple in 2022, I enhanced the photos memories experience by making it more personalized to the user.

At Nvidia in 2021, I created an oracle network to produce semantic segmentation pseudo-labels for previously unseen data to reduce overall datatset labeling costs.

At SRI International in 2020, I created baseline agents for the DARPA Machine Common Sense project.

Teaching

I've given guest lectures for the following courses in the Rutgers University ECE department:

  • Introduction to Robotics and Computer Vision - (14:332:472) - F'19 and F'21
  • Robotics and Humanity - (01:090:101) - S'24
  • Socially Cognizant Robotics - (16:332:590) - S'23 and S'24
Additionally, I've TA-ed for the following courses in the Rutgers ECE department:
  • Professionalism and Ethics - 14:332:393 - S'20
  • Probability and Random Processes - 14:332:226 - S'21