About Me

David Harper, PhD

I am David (Michael) Harper, a Mathematics Ph.D. graduate from Georgia Tech, class of 2023. My academic focus regards analyzing algorithms and optimization within random graph models. The hands-on experience of conducting simulations and analyzing data from these models fostered my strong interest in data science. This portfolio is a reflection of my expertise in the field. It showcases a variety of projects that demonstrate my skills in data science and analysis. Additionally, it highlights my familiarity with industry-standard tools and technologies, including Python, C#, QGIS, and SQL. I invite you to explore this portfolio to learn more about my background.

Projects

Auto-Burglaries Case Study

An in-depth case study on San Francisco auto-burglary trends before and after the pandemic.

Bibliometric Analysis and Topic Modeling of Probability Research

Bibliometric analysis of 20,000+ arXiv probability papers using BERTopic.

YouTube Comment Sentiment Analyzer

Flask app that analyzes YouTube comment sentiments using the Hugging Face API.

Kaggle Disaster Tweets - Sentiment Classification

A classifier for Tweets achieving top high performance.

Statistical Mechanics Models Library

A powerful python library for simulating statistical physics and random graph models.

C# Deep Learning Library

A custom C# deep learning library with ANNs, RL algorithms like Deep Q-Learning.

Web-Based Deep Q-Learning Demo

A web app for real-time experimentation with Deep Q-Learning.

Self-Driving Car - Imitation Learning

A unity based simulation of a self-driving car within a customizable 3D environment.

Pong Player Training Ground

A dynamic training platform for algorithms to play Pong against both human competitors and self-play scenarios.

Research

Figures and images of my work.

Overview

My research focuses on the relationship between randomness and algorithms with an eye for applications in machine learning and data science. I use methods from theoretical computer science and statistical physics to understand how small-scale interactions can influence larger system behaviors. I'm also interested in optimization within random network models. A significant part of my work in randomized algorithms considers what they can reveal about a system's noise sensitivity and robustness.

Publish Papers and Preprints

Transitions for exceptional times in dynamical first-passage percolation [Journal] [Preprint]

(with M. Damron, J. Hanson, and W.-K. Lam) Probability Theory and Related Fields

Non-optimality of invaded geodesics in 2d critical first-passage percolation [Journal] [Preprint]

(with M. Damron) In and Out of Equilibrium 3: Celebrating Vladas Sidoravicius

Upper bound on geodesic length in 2D critical first-passage percolation [Preprint]

(with E. Bates, M. Damron, X. Shen, and E. Sorensen)

Exceptional behavior in critical first-passage percolation and random sums [Preprint]

(with M. Damron, J. Hanson, and W.-K. Lam)

Blog

I'm passionate about articulating my interests through blogging. My posts span a spectrum from the practical to the theoretical, covering subjects like data science, probability, finance, and technology.

The Universal Approximation Theorem for Neural Networks

A discussion and proof for the Universal Approximation theorem for neural networks.

Discovering Trending Topics in Probability Theory with BERTopic

An analysis of research trends using NLP and clustering capabilities provided by BERTopic.

REST Assured: Simplifying Web APIs

An overview of REST API architecture and its practical applications.

Games

I also love creating games and I get particularly excited about incorporating mathematical elements into my games. Here, I showcase some of those games.

Enchanted Jackpot

A browser-based slots style game with a fantasy theme. Developed in the Unity game engine and written in C#.