Portfolio

Capstone

• Arria Boost: Sports Analytics Data Storytelling (Report)
  • Engineered a probabilistic machine learning model for NCAA Division I soccer rankings, surpassing professional soccer baseline predictions by 5%.
  • Validated model adaptability and enhancement across seasons through continuous learning.
  • Created a dynamic, user-interactive dashboard for real-time analytics, enhancing decision-making with comprehensive team performance insights.
  • Achieved 83.3% accuracy in men’s team field predictions by modeling match draws and tournament projections, outperforming FIFA rankings in predictive accuracy.
  • Feature Engineering, Statistical Machine Learning Modeling, Data Visualization and Pipeline Automation.

Statistical Modeling

• StockX Sneaker Analysis
  • Analyzed StockX 2019 data to identify factors influencing sneaker resale value, using R for multiple linear regression, highlighting significant variables affecting “hype value.”
• Job Training on Wages Analysis
  • Employed logistic regression to evaluate the impact of job training on wage increases, demonstrating higher likelihoods of wage improvements with training.

Cloud Data Engineering

• Strava Microservice - Continuous Delivery of FastAPI Data Engineering API on AWS
  • Developed a FastAPI microservice to deliver performance data visualizations via JSON, integrated CI/CD with GitHub Actions for seamless deployment on AWS.
• Strava CLI
  • Crafted a Python CLI for Strava activity data, utilizing Click for user-friendly interaction.
• Marathon Finish Predictor
  • Implemented a Flask-based service predicting marathon times using machine learning.
• Stroke Predictor Microservice
  • Created a Flask and sklearn application predicting stroke likelihood, containerized with Docker for easy deployment.
• Product Recommendation Engine

Unifying Data Science/Causal Inference

• The Legacy of the Taliban on Female Education in Pakistan (Report)
  • Investigated Taliban’s impact on female education through causal inference, revealing complex effects on school enrollment rates.

Machine Learning

• Model-Based Approach to Music Genre Assignment (Report // Presentation)
  • Developed machine learning models for music genre classification using Spotify data, improving genre assignment accuracy.

Mobile Application Development

• Mobile Passport
  • Developed mobile applications enhancing visitor experiences in NC state parks
• Duke Roster
  • Developed a mobile application that presents information about students within a course taken at Duke University.

Duke Data Plus

• Plus Explorations Project
  • Led a team in analyzing Data+, Code+, and CS+ program data over 8 years, enhancing decision-making and program efficacy with a comprehensive dashboard.

Algorithmic Trading - Financial Data and Modeling

• Powerful Pairs Trading
  • Applied K-Means Clustering for strategic pairs trading across sectors, demonstrating a novel approach to algorithmic trading.