VISHAL DUBEY
👋🏽 Hi, I'm Vishal!
I'm from Orlando, FL, and I'm building the future of joint banking with Ivella, an early-stage fintech startup. I am also currently on leave from Duke University where I was studying Computer Science. I'm really interested in driving impactful change through data and tech in the financial services and healthcare industries.
From my prior experiences at various biotech startups, I found a passion for entrepreneurship and impact, while also having grown my own technical software engineering skills. Last summer, I followed this passion while interning at UnitedHealth Group where I constructed a nationwide drug shortage platform for data visualization and recommendations. For more, check out my resume!
On campus, I'm part of the Executive team of the Duke Applied Machine Learning Group, where I help coordinate events and meetings for over 100 students, and am a Consulting Analyst at the Duke Impact Investing Group, where I research impact metrics for a disaster-preparedness NGO and an impact-focused Venture Fund. I'm also a part of other entrepreneurship and business-related groups as well such as the Student Founders program.
Fun fact about me: a friend and I once 'cracked' the Subway receipt cookie code algorithm in middle school and got a thousand free cookies 🍪. Some of my interests include sustainability, politics, martial arts, and gluten-free cooking. I'm also a fan of the Orlando Magic and Duke Basketball!
📸 My Life
My life and personal values are a product of my upbringing in an Indian household and certain role models who I've looked up to throughout. My personal role models are Jack Dorsey, Casey Neistat, and Satya Nadella. Together, they're some of the most creative minds and embrace my values as well!
In my free time, I love watching new TV shows (Billions is great right now), learning new gluten-free recipes, and playing strategy games like Catan. I'm also a huge fan of mexican and italian cuisine
Personal Values
- Ambition: I always try to know what I'm striving towards, even though it can be tough at times.
- Collaboration: constant and open communication are crucial to success and I always aim to carry them in any team I join.
- Accessibility: As someone who currently struggles accessibility issues, having equitable opportunities for all is very important to me, whether it's regarding product features or team contribution.
Music
I'm a huge fan of music and I can't code without it. I usually listen to around 5 hours every single day across all genres. My three favorite artists are Eminem, The Weeknd, and Drake but recently I've begun listening to more Anderson .Paak and Fleetwood Mac.
My appreciation for music began from a young age when I learned classical piano, and it continued as I picked up the clarinet and saxophone. Since, it's shaped how I approach the world from a creative standpoint and I've seen my tastes accordingly shift over time.
Here's a running playlist of my favorite songs ->
Some pictures I took
Duke Chapel - April 2019
First time seeing fall colors 🍁 - Nov. 2019
Florida beach and sunrise - July 2020
🛠 Some of my projects
🛠 Some projects
These are just a few highlighted projects which I loved building. For a more comprehensive view, email me or check out my GitHub!
Environmental Hazard Reporting App
February 2020
I led a team of 4 to create an automated system detect and report environmental hazards. The need for this app came as we observed litter, fallen trees, and cracked sidewalks in the streets of Arlington, VA. I built an iOS front-end app in Swift with a back-end RESTful API in Flask, and an administrator dashboard in React JS. One notable feature I implemented was web automation with convolutional neural networks on Google Cloud platform, allowing for the images taken to be processed and correctly sorted by hazard type.
This project was build for HoyaHacks 2020 at Georgetown University and ended up winning 2nd place. Following the competition, I began talks with Arlington City officials to implement this app within their Parks and Recreation department.
Technologies Used: Swift, Python, React, Google Cloud
Article Analysis Platform
October 2019
I built a platform to optimize the relevance rankings of news articles scraped from New York Times. This was done through building a RESTful API with a server-side multi-label classifier for NLP based on convolutional neural networks. My project ultimately won me the HackNC ‘Best Use of Data Analytics’ out of more than 100 teams.
Technologies Used: HTML, CSS, Python, AWS
NFL Fantasy Points Predictor Model
Aug. 2019 - Dec. 2019
As part of my Fall 2019 class "Machine Learning and Computational Modeling," I created a predictor model for fantasy points of NFL players in the past 20 years based on college season and NFL Combine statistics. To do this, I applied regularization and bootstrap regression in Scikit-learn using BeautifulSoup and Selenium for automated data scraping. I evaluated model efficacy using Mean Squared Error and also implemented PCA (Principal Component Analysis) in the process to select from over 100 input variables.
Technologies Used: Python, Scikit-learn, PyTorch
Surgical Simulation
May 2018 - July 2018
At the University of Florida during Summer 2018, I worked in the Computer Information, Science and Engineering Department under Dr. Jörg Peters to create a surgical simulation using programming in Python and Java. This research has far-reaching implications for surgeons and specialists who can use the platform to practice and create specialized scenarios as it can help train future generations of surgeons through virtual reality scenarios. In the last few months, my work has since been introduced into the UF database. Further, I presented my paper, “Developing a Surgical Simulation for a Laparoscopic Hepatectomy,” in front of a team of researchers and professors at the conclusion of my research project.
Technologies Used: Python, Blender, MatLab
💼 Work
TL;DR: I'm currently building Ivella, a fintech startup which specializes in joint banking for couples. Check out my LinkedIn for more!
Building the best joint-banking products for couples, starting with a new way for couples to split and manage their finances.
As a Fellow at the National Institute on Aging: Division of Behavioral and Social Research within the NIH, I researched hashing algorithms to conduct probabilistic linkages across multiple datasets.
Through my work, I assessed the effectiveness of multiple current hashing solutions, ranging from private contractors to internal SHA-512 Hashing Tools, and effectively matched error rates with external clients, all outlined in a report with my official results and recommendations as well as a presentation to executive members at the NIH.
Technologies Used: Python (Pandas, Scikit-learn, NLTK)
During my internship, I constructed a back-end environment to automate and centralize national drug shortage visualization, translating into 85% satisfaction rating from pharmacist testing and achieving a 0% error target within a month.
Specifically, I worked to impelement multiple APIs, spanning from geolocation through MapBox to querying drug prices from GoodRx. I also implemented support for OptumRx claims data within the visualization platform, increasing data visibility by 50%.
Technologies Used: MongoDB, Express, React, NodeJS, JavaScript, OpenShift Cloud Hosting
Duke Applied Machine Learning Group
EXECUTIVE BOARD,
MACHINE LEARNING ENGINEER
December 2019 - Present
TOURTech
(Dec. 2019 - May 2020)
Through DAML, I worked with TOURTech, an event services and concert routing company, In this project, I helped construct a platform for event-based network anomaly detection and prediction.
My contribution lied within bringing the two-tried solution based on machine learning and router log data into reality. This was done with a React web-app for parsing incoming logs and an API to process the data through an LSTM neural network with invariant clustering.
Technologies Used: Python (PyTorch, TensorFlow, Scikit-learn), React, JavaScript
Although I began as a research assistant, this research project transformed into a startup for nutrition recommendation, Nutrics. Within this health recommendation app, I worked to integrate with Apple Health data and Geolocation APIs, as well as engineer middleware to work with the server-side recommendation learning algorithms in TensorFlow.
I also helped gather data and eventually attract $75,000 in grants from the National Institutes of Health to continue the work.
Technologies Used: Swift, Flutter, Python (Selenium, BeautifulSoup, TensorFlow)