I solve real world problems with machine learning.
Currently
Internship at IBM Research, May – Aug ’22.
AI Foundation Model training, dataflow optimization. Based in historic Yorktown, NY.
Masters in Computer Science from UIUC
Specialization in applied machine learning.
Expected grad May, 2023.
National Center for Supercomputing Applications (NCSA)
Research Assistant, Oct 21-Present.
Funded by the NSF & IBM Research.
I implemented distributed ML training on a GPU supercomputer (8 Nvidia DGX nodes), enabled by domain experts in biology and physics.
Previously
Sarcos Robotics – Advanced exoskeletons
Machine Learning R&D Intern
Built machine learning vision segmentation models running on edge computers (Nvidia Jetson on-robot) including sensor fusion with depth sensing cameras to understand 3D scenes in pixel-per-pixel segmentation, at 100+ FPS, for highly accurate robot planning.
Previous cofounder of Bioloop Sleep
Jason Jin has built a great team since I left!
View my resume here.
Tools for Thought
🧠 Neural networks ignoring training data: the astonishing power of Attention
Learning means forming a compact representation of a concept and forgetting enables greater information density.
📚 Designing Better Books
What is the future of books? How can we make learning cheaper and easier by design? Like with a mnemonic medium? How can I build on the best examples on the internet today?
🛠 My best tools as a high-performance knowledge worker
Pianists at every level drill scales so they can think in higher abstractions while they play. But no knowledge worker I know deliberately practices their craft. I’m hoping to bridge this gap, to craft a nice workspace, to leave more mental space free to invent new ideas.
Come look at the “tech stack” behind my creative process.
AI Research
🥇 How We Won Argonne Labs’ AI-Hackathon
Novel AI for Material Science Problems
🧠 What AI algorithms win in the real world?
A Theoretical Analysis of the Algorithms that Win Kaggle Competitions
😏 Unsupervised Facial Recognition: an Extension to InfoGAN Generative Adversarial Networks
📰 Fake News Detection innovation using “Stance” (e.g. agree/disagree text corpus classification)
⌚ Using Wearables for Sleep Improvement: A Platform for Randomized Self-Experiments [Preprint]
The Good Thinking blog
💻 Making Windows Excellent (become a 10x computer user)
💡 Most enlightening places on the internet (for CS nerds)
🍎 Best Apps for iOS (from a life-long android fanboy)
🎓 Academic Publishing is a joke; here’s a better way to communicate science. Draft, please add your comments 🙂
My work
NASA In-Space Robotics Internships
Won HackMIT 2017 & 2018!
Climbing First Ascent using Custom UAVs
See the photo & video here!
NASA Media Highlights
Follow my twitter
Follow my twitter 🐣 for machine learning news & memes.
Personal Life
Kastan Day 