Arsh Banerjee

CS @ Princeton University

Fun Facts 🥳: 



The intersection of science and technology has always fascinated me, and this passion has guided my academic journey at Princeton. In my senior year, I am pursuing minors in Quantitative Decision Science, Statistics and Machine Learning, and Cognitive Science. My internships and research opportunities have been vital stepping stones in helping me explore Computer Vision, NLP, and how to use technology to solve real-world problems. I am always excited for any opportunities to conduct research or roles that help make an impact.


  • Detecting AI-Generated Images Created by Diffusion models

    This project involves the development of a tool for detecting AI-generated images, specifically from diffusion models, to help counter misinformation and qualitatively understand identifying key image attributes for this class of classification task.

    • Computer Vision
    • Generative AI
    • Tensorflow
    • GANs
  • Robotic Path Planning

    A proof of concept project completed as part of my Verizon internship in the Summer of 2022. The system tracks multiple robots and changes their velocities autonomously. The project was used to demo the low-latency capabilities of 5G edge computing via AWS and Verizon's 5G network in the automotive space.

    • Robotics
    • ROS 1 & 2
    • Python
    • Computer Vision
    • Spatial Mapping
    • AWS
    • Cloud Computing
  • Image Geolocation with Computer Vision

    This project aimed to create a system to perform geo-location, which involves predicting the location of an image using only pixel data, by implementing and evaluating differing model architectures and model types.

    • Computer Vision
    • Tensorflow
    • CNN
    • Transfer Learning
  • Unsupervised Discovery of Textual Implicit Gender Bias: A New Analysis of Reddit and Fitocracy

    A group project completed alongside Pierce Maloney and Christian Ronda for Princeton's NLP Course. We implement a causal framework established by Field et al. to identify implicit gender bias at the comment level in two corpora: Reddit and Fitocracy. Our work offers insight into how implicit gender bias detection can differ across different social platforms.

    • NLP
    • Unsupervised Learning
    • Sentiment Analysis
    • GANs
    • Adversarial Training