loreLike every lore, there's always more.
Below is centered around baseball.
Came back to school to wrap up my studies this winter.
I built OpenCommand, the world's most accurate throwing accuracy tracker using computer vision & mixed models.
It beat every human-based trackers, and was covered by The Athletic (NYT), Sportsnet, FanGraphs, and Baseball Prospectus.
I was an Associate Quantitative Analyst for the Philadelphia Phillies, where I was part of the predictive modelling team, making out probability models.
I was an Associate Data Scientist for the Baseball Development Group, where I did baseball research and built ML models.
I also did some software engineering. I revamped the athlete assessment package to include things like balance tests, reaction speeds and neurotransmitter profiles.
Took a gap year to immerse myself in philosophy and get out of misery. Best decision ever!
I then went to University of Toronto to learn coding.
I worked at Swiss Chalet to pay for my flight to Louisiana where the top baseball training facility was.
This was the worst time of my life.
pet projectsSee my up to date projects on
GitHub/Twitter.
The Intent to Give it Your All was an attempt at demonstrating sports science concepts1force-velocity curve, stretch-shortening cycle using ML.
It measures sprinting effort by taking the ratio between exerted force vs maximum force at each timestamp, for each player.
RenderMRI makes it super easy to visualise MRI imaging, which is hard for non-professionals.
It takes images from multiple directions, uses segmentation to trace organs, then constructs a 3D object using polygons that match the traces.
This video is a 4D reconstruction of a throw using Gaussian splatting.
It only uses 8 cameras2Gaussian splatting typically involves 30+ cameras, and makes up for it by building one skeleton for the whole throw and posing it in every frame.
Shootout.app was a web game that I made out of spite after hearing that coolmathgames.com generated 8 figures.
Your goal is to look at the penalty kicker's face and guess if they'll score. The game hit the world cup wave and went super viral! (100k+ plays)3No one played twice, though. (Turns out it's all luck and no skill.)
misc
- Come watch me train
- 내일을 즐기려면 오늘도 즐겨야해. 난 네가 오늘을 즐겼으면 좋겠어.