Eshan Arora

I am a grad student at Northeastern University, Boston, MA where I focus on key areas such as machine learning, algorithms, and statistics. I have garnered a strong foundation in Machine Learning, Data Science and Statistical modelling. Through my courses, research and projects, I’ve learned to effectively analyze data and use it to derive meaningful conclusions.

Research Experience

My current research involes developing and implementing machine learning models, specifically for predicting discrete and waveform features from the challenging ICU healthcare datasets, MIMIC III and IV.

During Summer 2023, my association with Helping Hand’s Lab by Dr. Rob Platt brought me closer to the innovative realm of reinforcement learning. Here, I contributed to a novel paper, delving into symmetric models for visual force policy learning. The essence of my research also resonated in programming the baselines of the Visual-Tactile Transformer, which aimed to offer a multi-modal fusion platform for state-of-the-art robotic manipulation.

Fall 2020, I showcased a solution to colorization of grayscale images to gain deep insights on the factors that influence the coloring process using Generative Adversarial Networks (GANs), advised by Ass. Prof. R.P. Gohil of Computer Engineering Department.

In Spring 2019, my team worked on generating Audio Samples using Generative Adversarial Networks advised by Prof. Dr. Kishor D. Uplah of the Electronics Engineering Department. Unfortunately, due to the COVID-19 pandemic, this project is still in progress.

Work Experience

Previously, I worked at Deloitte India (Offices of the US). I have extensively worked on the ServiceNow platform for automation over the course of many projects where I have implemented IT service management (ITSM) modules, UI Actions, testing via workspace and executing client-side modules (UI Policies, Client scripts) and server-side modules (Business Rules).

In Summer 2021, I completed the Google Summer of Code journey under the umbrella organization The Python Software Foundation. I engineered a robust automation solution flr’s product Deeplake: a one-stop solution for streaming petabyte-scale datasets without sacificng performance. My contributions such as ingest and ingest-kaggle have automated the ML datasset ingestion process.

A prime example of my work is that ingesting the CIFAR-10 dataset on Deep Lake now takes 96% less code. What earlier took several hours of boilerplate code, now takes just 1 line of code. I worked on other utility features such as “Auto-detection of dataset compression” and “Post-ingestion summary of dataset” which further enhances Deep Lake’s capabilities.

What do I do when I don’t do

I am always down for a game of Lawn Tennis :tennis: and I can’t get enough of Photography, check out my clicks here :camera:

I harbor an interest for Instrumental music and my favorite artist is John Mayer :musical_note: