M I C H A E L
machine learning
00
machine learning · virginia tech

MICHAEL OWEN

I build intelligent systems, from optimizers written
from scratch to real-time computer vision.

currently Data Analyst @ BodBx
studying MS Computer Science ’27
based in Blacksburg, VA
scroll
01 about

I'm a computer science student at Virginia Tech turning data into decisions by training models on athlete kinematics, rebuilding optimizers from first principles, and shipping the interfaces that make ML legible.

5+

years writing code, from C++ to production TypeScript

90%+

validation accuracy on a gesture CNN I designed & trained

+35%

accuracy gained through ensembling in the Kaggle AI Math Olympiad

2

degrees in progress: BS ’26 & MS ’27, Computer Science

02 skills

The stack I think in.

a.

Machine Learning

My core focus: model design, training & evaluation

  • PyTorch
  • TensorFlow
  • scikit-learn
  • Pandas
  • NumPy
  • OpenCV
  • MediaPipe
  • Jupyter
b.

Languages

From systems programming to typed frontends

  • Python
  • TypeScript
  • JavaScript
  • Java
  • C/C++
  • SQL
c.

Web & Frameworks

Where the models meet the user

  • React
  • Next.js
  • Express.js
  • Tailwind
  • Netlify
  • Git
03 selected work

Things I've built.

001

ML Playground

An interactive ML textbook covering six algorithms, with OLS, batch gradient descent, SGD and Adam implemented from scratch, recording per-epoch trajectories you can replay step by step on a live loss chart.

// built with
  • Next.js
  • React
  • TypeScript
  • Tailwind
  • SVG viz
002

MichaelsMusic

A custom Spotify web player on the Web Playback SDK, with real-time streaming and playback control, an Express.js serverless backend handling OAuth and API proxying, deployed on Netlify.

// built with
  • JavaScript
  • Express.js
  • Spotify API
  • OAuth
  • Netlify
003

Gesture Recognition

Real-time hand gesture recognition: a neural net with batch normalization and dropout at 90%+ validation accuracy, fed by an augmentation pipeline that 5×’d the training set. Live webcam training for 8 gesture classes.

// built with
  • Python
  • TensorFlow
  • OpenCV
  • MediaPipe
  • CNN
004

Kaggle AI Math Olympiad

Built ML solutions for olympiad-level math in the AI Math Olympiad. Implemented and optimized three PyTorch models, improving accuracy 35% through ensembling and hyperparameter tuning.

// built with
  • PyTorch
  • Ensembles
  • Feature engineering
  • Tuning
see everything on github ↗
04 experience

Where I've been.

2026 to now

Data Analyst · BodBx

Gathering kinematic data from Virginia Tech athletes with the Strength & Conditioning program, applying supervised and unsupervised learning to surface the patterns connecting weight-room training to competition results.

2020 to 2025

Athletics Manager · Highlands Swim & Tennis

Ran logistics for 20+ competitions a year and coached 30+ youth athletes. Five years of leading teams, building training plans, and shipping on schedule.

2022 to 2027

Virginia Tech · BS + MS Computer Science

Bachelor's expected May 2026, accelerated Master's expected May 2027. Focus on machine learning and data-driven systems.

05 contact

Let's build something intelligent.