I’m an Electrical and Computer Engineering graduate student at the University of Georgia, which is a bit of a pivot from my Statistics and Finance undergraduate degrees. For as long as I can remember, I’ve loved mathematics and have always found interest in forecasting and building models, and I’m particularly fascinated by integrating multiple elements of data analysis such as spatial-temporal models; I aim to complete my GIS Certificate at UGA with coursework in deep learning for geospatial analysis. Currently, I am a research assistant at the UGA Multispectral Imaging Lab (MILAB) and at the Controlled Environment Agriculture (CEA) Lab. I have been recognized for my outstanding research in biometrics, where I implemented and reviewed the effficacy of various deep neural network architectures on varied recognition challenges, and in horticulture, where I am working with localized nutrient deficiency detection and open-source software development. Additionally, I have a strong understanding of market forces and financial markets from my education in fixed income securities, derivates, options, and futures markets, and my underwriting and virtual prospecting expperience as an analyst at a real estate private equity firm.
See my resume here: Resume
What I do
I got my start in statistics when I took AP Statistics in sophomore year of high school, and my teacher, Mr. Crowers, integrated current events (i.e. sports and political primaries) into classroom discussion, and showcased the interactive data visualizations found on FiveThirtyEight. I loved playing around with the simulations to view how they would be impacted by imagined scenarios, and having completed AP Statistics left me wanting to learn more. My first statistics class in college involved learning R alongside ANOVA tables and logistic regression, igniting a newfound love for programming and data visualization. During my sophomore year, I took further classes in statistical programming and regression analysis; while I was strengthening these fundamental skills, I wanted to learn more about advanced modeling techniques. I found the carykh YouTube channel, where Cary Huang posts videos covering topics in unsupervised learning, deep learning, and evolutionary algorithms, which ignited my desire to pursue machine learning. Over the 2020 summer, my internship focused around the evaluation of a convolutional neural network model trading foreign exchange futures; while at the time my understanding of the model was quite limited as the structure was quite sophisticated, I kept returning to the articles underlying the model’s design in order to gauge my understanding of the process. In the fall of 2020, I took courses in time series analysis and statistical programming with R and Python; the former invigorated a desire to learn about how to model volatility in financial markets, and the latter got me comfortable enough to begin learning more advanced packages such as R’s Shiny interactive data visualization package, and TensorFlow. I spent the beginning of 2021 learning about machine learning through a playlist of Texas A&M Applied Multivariate Analysis lectures which made me dead-set on learning these skills in a classroom environment.
While balancing being a full-time student and interning at 33 Holdings, a private equity real estate investment firm, I entered into UGA’s 2021 Data Science Competition, where I compared the classification capabilities of random forest, XGBoost, and multilayer perceptron to that of a logistic regression model, applied to a highly-unbalanced classification problem of detecting which customers will default on their credit card due to nonpayment. Although I did not win the competition, I felt so much satisfaction from the dedication I put in and the knowledge I learned that I made it a priority to further build these skills and take classes, both online and through UGA. As a result, I chose to take Pattern Recognition (a course in traditional ML techniques) and Deep Learning, two sequential classes in the College of Computer Engineering despite those classes not counting towards progression of my degree requirements. Last fall, my semester-long project for Pattern Recognition involved the classification of white blood cell nuclei as either leukemic or healthy; although my project did not generate satisfactory results (expected due to the specific avoidance of using neural networks), it laid the groundwork for future work in my Deep Learning class when I revisited the problem. With fewer than ten days left in the semester, I focused all of my efforts onto my Network Data and Graphical Models class final project, where I classified article headlines as either coming from a genuine or fake source using a character-level approach.
In addition to my in-person classes I have completed Bloomberg Market Concepts, which covers financial markets, fixed-income securities, forex, and the use of Bloomberg Terminal, and the IBM Machine Learning Professional Certificate, which covers supervised regression and classification techniques, unsupervised learning, time series analysis, and deep learning. In the future I aim to take courses for SQL for data science, Shiny for data visualization, advanced TensorFlow model building, and MLOps, and complete Google’s TensorFlow Developer certificate program. In terms of potential areas I’d love to learn more about, I am interested in sharpening my understanding of graph-based neural networks, geospatial data, information systems, and implementation of real-time data in machine learning.