Justin L. Kim
Wrangling data, breaking open black boxes, and building software to last.
About Me
Justin Kim here, I'm a software developer, researcher, and author. Between working in software development in industry, and building machine learning models in academic settings, I've built up an expertise in working with complex codebases and gained an appetite for making sense of the unknown through data.
My main strengths are in developing machine learning frameworks in Python, and I hope to grow my experience applying these skills to code in production environments. More than anything, I love a good challenge and biting off more than I can chew. Pressure makes diamonds!
Background
Experience
Research Assistant
USCF Center for Intelligent Imaging • 2025 - Present
San Francisco, CA
San Francisco, CA
PythonTensorFlowComputer VisionLLMs
Analyst Developer Intern
Boeing • Summer 2024 • Huntington Beach, CA
JavaDatabasesSecurity
Publications
Independently Published (Amazon) • 2024 • ISBN 979-8333684905
An introductory Statistics, Python, and Machine Learning textbook with examples using Scikit-Learn, Facebook Prophet, and XGBoost to compare statistical models like Linear Regression & ARIMA to machine learning models like Gradient Boosting.
Synthesized forecasting examples & images into a 27,000-word textbook with editorial support.
Synthesized forecasting examples & images into a 27,000-word textbook with editorial support.
Justin Kim, Kyoungwan Woo — GLEAN: Generative Learning for Eliminating Adversarial Noise (Pre-print)
Pre-Print • 2024 • DOI: 10.48550/arXiv.2409.10578
Developed an image-to-image GAN framework to bypass Glaze (an adversarial perturbation generator), revealing vulnerabilities in AI artwork-protection methods.
Cleaned and organized 261 GB of tabulated data in Pandas to train a 32k-parameter GAN using Fast Fourier Convolutions and ResNet blocks in TensorFlow; trained on 8×H100 GPUs on MIT HPC with annealing LR.
Cleaned and organized 261 GB of tabulated data in Pandas to train a 32k-parameter GAN using Fast Fourier Convolutions and ResNet blocks in TensorFlow; trained on 8×H100 GPUs on MIT HPC with annealing LR.
Education
B.S., Electrical Engineering / Computer Science
University of California, Berkeley • Currently Enrolled
Computer ProgrammingML