Projects

(2024) Parlaid, a full-stack web application built with React, Typescript, Next.js, Flask, and SQLite, assists sports bettors in creating strategic NBA parlays. By leveraging historical player performance data and betting lines from the 2024-2025 NBA season, the platform uses a proprietary neural network that processes player statistics, team dynamics, and contextual game information to identify correlations between over/under prop bets and suggest high-probability combinations for user-defined inputs.

In collaboration with Kyle Shepherd and Sunil Rosen, we produced valuable insights into prop correlations. Dive deeper by reading our Blog Post on Medium.

(2024) Built a browser extension that won Runner Up at Nasdaq's Americas 2024 Hackathon, which harnesses your tab's webpage content to complete various tasks. The assistant successfully:

• Streams out a ready-to-send reply in your inbox.

• Enhances code readability with comments.

• Searches for company CFOs and generates tailored LinkedIn sales pitches.

"AI Browser Assistant" was developed with React and Flask using AWS Bedrock, Elasticsearch VectorDatabase, and SQLite.

(2023) Using React for the frontend and Flask for the backend, I developed a chatbot that mimics ChatGPT's website style, with the help of OpenAI's API. This project is further enhanced by integrating an SQLite database to store and display conversation histories on the sidebar, sorted by the date of the latest message in each conversation. 

Additionally, I implemented text analysis, enabling users to view a histogram of the most common keywords in chat queries. Users can click on a specific keyword, triggering a table display of conversation data where that keyword appears.

(2023) Leading a small team, we leveraged the tidyverse and dplyr libraries in R to meticulously wrangle data from various websites, aiming to assess the impact of NIL on D1 College Football. I built and developed a comprehensive Shiny application including various interactive elements, notably a Leaflet map that visualizes NIL's effects on the transfer portal. 

It also includes a logistic regression model predicting transfer portal probabilities, utilizing quantitative and categorical predictors such as athlete rating and position. You can explore more features and insights by visiting our project here. 

(2023) Working alongside fellow Amherst College students, we developed a website featuring an Express backend and integrated PostgreSQL. This platform is designed to efficiently visualize, search, insert, and delete data from the extensive collection of over 30,000 specimens curated by the Beneski Museum. Due to the sensitive nature of the data, I cannot display specific examples publicly.

(2021) "LogChecker" is a Python automation tool developed for ~70 contractors and staff members used over 20,000 times. The tool fully supports five game modes across two servers and features permissions-based access tailored to various staff ranks. Key functionalities included a redo feature for executing previous commands with new log files and a leaderboard connected to an SQL Database to track usage and tally scores.