Mason Choi

Hi! My name is Mason Choi. I'm a current senior pursuing a dual degree in computer science and math (stats+ds concentration) at the University of Massachusetts Amherst. I've worked in the finance and real estate industries as a data engineer and am now exploring the fashion AI startup space. My interests lie in computing for the common good, particularly in scalable cloud infrastructure and AI safety. I'm a competitive fencer at the national level, and an avid options gambler, 3d printer, climber, and fisherman 🎣.

Projects

These are some of the projects I have developed over the years, starting from the most recent to the oldest. I have been creating projects with code since 2018.

Campus Quad

November 2023 - May 2024

A mobile app developed under parent company QUAD, seeking to bring college students together under one campus. The app features both a social and marketplace side hosting features ranging from chat rooms based on academic class, forum posts for news and sports, and buy, sell, and rent options for goods and services.

Lexia

March 2023

A simple Google Chrome extension written in javascript to improve the readability of online text, made specifically for a friend with dyslexia.

Albert

January 2023 - March 2023

Albert is an experimental research project exploring the possibilities of using a transformer model such as GPT-3 to help train a custom transformer model, Albert.

OpenAI Text to Notes Converter

December 2022 - January 2023

A simple website hosting custom technology made possible by OpenAI's GPT-3 API. The technology can convert any textbook text into note form broken down by topic, effectively streamlining the notetaking process.

Fantasy Fencing X The Fencing Project

September 2022 - Archived

A small dev team founded by fencing celebrity Cyrus of Chaos to deliver a Fantasy Football inspired fencing platform. Incorporates elements from my project "The Fencing Project" which scrapes and manipulates all available fencing data. I mainly head backend development and systems design, occasionally touching our beautiful frontend UI.

Melting Point Estimator

July 2022 - September 2022

A deep learning neural network built using Tensorflow in Python which predicts the melting points of molecules given their molecular structures in the standard SMILE notation and average mass.

Tunabl

July 2022 - Archived

This is Tunabl, a mobile application built on the React Native framework for discovering new music. Tunabl is still in early development with no estimated date of release.

Raytracer

June 2022 - July 2022

Custom raytracer built from scratch in javascript. In collaboration with Tony Zhang

Filess

March 2022 - July 2022

Filess is a static asset hosting service with the intent of seamless sharing made unique. Quickly upload images, videos, etc. and create your own custom URL link that can be shared. The website was built on top of the Flask framework with AWS S3 (cloud storage) file hosting.

Neural Network

April 2022 - May 2022

Built a neural network from scratch by implementing statistical classification methods in Python with NumPy. Ranks 6th best performing on the EMNIST Balanced benchmark dataset with over 96% accuracy.

Image Analyzation App

August 2021 - September 2021

React Native mobile application built with Expo, providing various tools centered around images. Contained Google's "Mobile Net" Convolutional Neural Network (CNN) model for classification on 1,000 classes.

My Machine Learning Journey

March 2021 - Present

I expanded on my Python neural network from scratch into a pseudo-library for machine learning in Python, Java, and C++.

Anchor was a website providing multiple regression model predictions on over 2,500 stock symbols. Its Tensorflow model was run daily in an AWS Elastic Cloud Compute (EC2) instance, assets stored in AWS S3, and website built with Django.

Simple fencing analysis tool that spliced fencing video footage for fencing frames only with a CNN, reducing average watch time by over 40%.

Both my AI composer and AI singer fashioned Long Short-Term Memory (LSTM) models to generate their own, original musical pieces. This required webscraping, data cleaning, and, of course, machine learning (ML).

You can view more of my projects on my Github

Work

Stealth Startup

XX XX - Present

Cofounded startup operating in the fashion AI space

Working on bringing exciting technology to the consumer markets in 2025!

AlphaSeeker AI

April 2024 - November 2024

Configured company servers to handle large streams of financial data

Deployed machine learning models to extract insights from them

Atlantic Prime Estates

July 2023 - April 2024

Built advanced housing data aggregation pipelines to push exclusive API data into AWS hosted MySQL databases for later mathematical analysis

Compared multiple regression and time series models to more accurately predict appreciation rates, building out the fund’s most valuable market analysis tool

Solve Advisors Inc.

March 2022 - July 2023

Worked with the AI/ML team to build a Business Development Company (BDC) financial statement parser, which disrupted the industry and saved the company an estimated 848 hours annually

Automated manual bond and loan data pricing with Selenium and Google’s Oauth API in an AWS EC2 instance

Built tools to aggregate global treasury yeild data and organize it

Advantage Data Inc. + Best Credit Management Inc.

September 2021 - March 2022

Developed Python scripts to scrape data from the SEC, parse and transform the data, and load it into the company database (ETL process), replacing hundreds of hours of manual work

Collaborated with dev team to deploy data pipelines directing source data into our system with Python and SQL