Education
Occidental College
Bachelor of Science in Computer Science, Mathematics, Physics
Aug 2023 - May 2027
Los Angeles, CA
Portland State University
Concurrent Enrollment During High School
Sep 2020 - Jun 2023
Portland, OR
Publications
Eye biometry prediction from ultrasound images using convolutional neural networks
John Lipor, Hadi Khazaei, Danesh Khazaei, Andrew Franck, John Ng, Faryar Etesami • 2023
Experience
Lipor Lab
Research Assistant
- • Co-author of research publication, "Eye biometry prediction from ultrasound images using convolutional neural networks."
- • Developed a baseline multi-output convolutional neural network to analyze ocular ultrasound images, with shared and independent layers based on GoogLeNet model
- • Leveraged image augmentation to inflate dataset size 150%
- • Supports novel neural networks method in diagnostic ultrasound
Team Strand Game Studio
Lead Game Developer
- • Winner of Regional Oregon Game Project Challenge
- • Led development of a Unity-based, cooperative, mountain climbing video game, Roped Together
- • Developed custom physics engine for 3D rope model, employed Blender for models/animations
- • Used Git and GitKraken for version control and weekly sprints
- • Mentored teammates in Unity, C#, and Git; delegated tasks and resolved conflicts
Projects
Machine Learned Detection of a Baseball Pitcher's Tell
Implemented several deep learning models for baseball pitch video analysis (ConvLSTM, I3D, Two-Stream) with up to 82% validation binary classification accuracy. High-dimensional calculus and linear algebra used to display the model's methods for prediction. Worked with SCIAC college baseball team to obtain a 250+ element video training set.
Modeling Neural Action Potentials w/ Hodkin-Huxley Model
Investigated the Hodgkin-Huxley model of neural action potentials. Model accurately reproduced key neurophysiological phenomena, including action potential generation, threshold behavior, and spike train dynamics. Developed multiple electronic blocking methods to simulate the effects of anesthetics on ion channels.
Custom Deep Learning Library in NumPy
Used Python/NumPy to develop custom deep learning library "MyTorch", which operates and calls functions analogously to PyTorch. Developed custom MyTorch loss functions, optimizers using stochastic gradient descent, and forward/backward passes. Created functioning convolutional, recurrent, and regular neural networks.
Forecasting Geothermal Power Plant Feasibility with ML
Developed ML model in PyTorch to predict geothermal heat flow residuals based on 28 feature dataset. Updated project to analyze topographical maps with convolutional neural network model. Accurately predicted geothermal heat flow residuals with a 77% binary classification accuracy. Utilized Pandas and Seaborn libraries to analyze and format professional dataset provided by the United States Geological Survey.
Academic Background
Undergraduate Coursework: Data Structures, Algorithms, Vector Calculus, Advanced Linear Algebra, Discrete Math, Abstract Algebra/Group Theory, Differential Equations, MATLAB Programming/Numerical Methods, Data Modeling with SQL, Information Theory, Dynamics, Math Methods in Physics, Physics/Dynamics Simulations
Graduate Coursework: Mathematics of Machine Learning, Theory of Deep Learning, Numerical Optimization
Technical Skills
Languages
Frameworks
Tools
Athletics
NCAA Baseball
Catcher, Occidental College Baseball Team
- • Balanced rigorous academic schedule with NCAA athletics
- • Developed leadership and team collaboration skills
- • Applied knowledge of ML to analyze baseball pitch mechanics
Contact Information
Email: franck@oxy.edu
GitHub: github.com/ajfranck
LinkedIn: linkedin.com/in/andyfranck
Location: Portland, OR | Los Angeles, CA