Hello, I'm
An enthusiastic Computer Science student, looking for opportunities in the field of Software Development, Data Science and Machine Learning.
ABOUT ME
My name is Joseph Anurag Mohanty. I am currently pursuing my Masters in Computer Science at Rice University, with a Bachelor’s degree in Computer Science Engineering from Manipal Institute of Technology. I am specializing in the field of Data Science and Machine Learning. With a strong foundation in software development and a keen interest in emerging technologies, I have continually sought out opportunities to apply my skills to real-world challenges.
Currently, I am working as a Research Assistant under Professor Chris Jermaine. We are working on a Machine Learning system - Einsummable - that automatically distributes ML compute graphs across clusters of CPU or GPU machines. It is based on the idea that tensors are relations mapping positions to values and all operations are therefore relational. The benefit of treating ML compute graphs as relational queries is that they can be easily reasoned and optimized over.
Prior to joining Rice, I worked as Tech Engineer at UBS in their Investment Banking Division, where I was part of the CAIP (Corporate Actions and Income Processing) Team. I primarily contributed to leading code release activities for our application suite.
PROJECTS
URL Shortener
Developed a URL shortener service using Go, MongoDB, and Gorilla Mux, allowing users to shorten, manage, and retrieve URLs efficiently through a RESTful API.
Brain Tumor Detection
Developed a custom CNN model to detect brain tumors with a test accuracy of 95%, enhanced by implementing GRAD-CAM for heatmap visualizations to highlight tumor regions.
Image Colorization using user-guided Hints
Implemented an AutoEncoder to colorize grayscale images, also added user-guided color priors for users to generate a variety of colorized images.
Sentiment Analysis of Complaints on a Student Grievance Portal
Engineered a sentiment analysis model using LSTMs to prioritize student grievances, achieving an 87% test accuracy and significantly improving response times to critical issues.
Intelligent Query Optimization
This project explores Dynamic Bloom Filters (DBFs) and the fine-tuning of SentenceTransformer models for optimizing semantic search in large-scale document retrieval systems. It focuses on improving scalability, precision, memory efficiency, and query performance in dynamic and real-time workloads.
PUBLICATIONS
Multimodal Sentiment Analysis of #MeToo Tweets
Published in IEEE BigMM Conference, 2020
Authors: Soham Tiwari, Priyam Basu, Joseph Mohanty, Sayantan Karmakar
Abstract: The #MeToo trend has led to people talking about personal experiences of harassment more openly. This work attempts to aggregate such experiences of sexual abuse to facilitate a better understanding of social media constructs and to bring about social change. We propose an approach to multimodal sentiment analysis using deep neural networks combining visual analysis and natural language processing. We have made use of a Multimodal Bi-Transformer (MMBT) model which combines both image and text features to produce an optimal prediction of a tweet's stand and sentiments on the #MeToo campaign.