Work Examples
Sentiment Analysis Trading Strategy
Developed and backtested a quantitative trading strategy that leverages Natural Language Processing (NLP) to capitalize on market sentiment. The project involved an automated Python pipeline that sourced daily news headlines for a universe of over 100 stocks using pygooglenews, performed sentiment analysis on each headline with NLTK's VADER model, and aggregated these scores monthly. A long-only portfolio was systematically constructed by selecting and rebalancing into the top 5 stocks with the highest positive sentiment each month. The entire backtesting framework, from data acquisition (yfinance) and cleaning to performance calculation and visualization, was built using Python with Pandas, NumPy, and Matplotlib.
Stock Watchlist
Created a Python Flask web application that allows users to build and manage a personalized stock watchlist. The app leverages the yfinance library to fetch real-time stock prices and key financial data, providing up-to-date information for each ticker in the user's list. Users can add or remove stocks, view live updates, and customize their watchlist according to their interests. All user data and preferences are stored in a local JSON file, ensuring persistence across sessions without the need for a database. This project demonstrates seamless integration of Flask for backend logic, yfinance for data retrieval, and JSON for lightweight data storage.
Genetic Algorithm for Block Game
Developed an AI agent using a genetic algorithm to discover optimal strategies for a custom block placement puzzle game. The system simulates thousands of agents per generation, evolving their decision-making logic based on a fitness function that rewards efficient and high-scoring solutions. Key features include automated gameplay simulation, dynamic mutation and crossover, and a real-time visualizer to observe the evolution of strategies. Made in Python, using pygame for interactive visualization, matplotlib for analytics and progress tracking, and numpy for efficient numerical operations.
Portfolio Website
You are here! Engineered and deployed a dynamic, fully responsive personal portfolio website from the ground up using Next.js and React, with a focus on creating a clean, modern user experience. Key functionalities include a theme-aware UI with seamless light/dark mode toggling, a secure contact form that leverages a custom API route for email delivery, and an interactive projects gallery with real-time search and filtering capabilities. The front-end was built with a custom library of reusable React components and styled with Tailwind CSS, ensuring a consistent and maintainable design system. The entire application is deployed on Vercel, utilizing its CI/CD pipeline for automated builds, hosting, and analytics.
Algorithms for Julia Sets
Developed both 2D and 3D visualizers for Julia sets and related fractals, utilizing rasterization and raymarching techniques, respectively. The project focuses on delivering an interactive and educational platform that allows users to explore the intricate mathematical properties of fractals in real time. Features include adjustable parameters for set generation, real-time feedback on escape rates, and visual explanations of the underlying mathematical concepts. Made with Unity with C# for the core logic and custom shader programming. This tool serves as both a learning resource and a demonstration of optimized rendering strategies for complex mathematical structures. Made for Purdue Experimental Mathematics Lab 2025 (PXML).
DeepPhish
Worked as part of a team to develop DeepPhish, an advanced cybersecurity consulting tool designed to predict and test phishing attacks. The project involved creating a sophisticated classification model to analyze and predict user behavior in response to phishing attempts based on their details and tech literacy, using machine learning and data analytics. Using these results, the tool utilizes self-hosted LLMs to generate personalized phishing emails, with varying difficulty levels, to simulate real-world phishing attacks and assess employee responses. An automated reporting system then generates a detailed PDF with insight into individual employees and provides recommendations. Made in 24 hours for Catapult Hacks.
