Artwork inspired by Piano Sessions: Reflections (Xeuphoria × OneChilledPanda); original cover art by wacca_わっか.
I’m studying Statistics & Data Science and Applied Mathematics at UCLA. I’m interested in machine learning, computer vision, and statistical inference.
- 🌱 Currently learning: transformers and diffusion models.
- 💻 Experience: Python, R, C++, and Java, plus front-end development with HTML, CSS, and JavaScript for my portfolio website.
- 🤝 Open to collaborating on: machine learning experiments, statistical computing, and open-source tools.
- 🏅 Competitive programming: USACO Gold Division and ACSL National Champion (two seasons).
- BruinML · CLIP and computer vision — Studying training-free detection of small unfamiliar image regions with a five-person team. My work includes Python simulations across 225 combinations of noise distributions, object geometries, and signal strengths to compare statistical detection methods.
- ACM AI @ UCLA · poker modeling — Developing a transformer-based poker model with teammates to study player behavior from sequences of game states and actions. In progress.
- 字生 (Zìshēng) · conditional diffusion — A runnable PyTorch pilot for my STATS 100A honors study, generating images of 12 Chinese characters with a class-conditioned diffusion model. Includes a saved checkpoint, sample gallery, and interactive reverse-diffusion sampler.
- PDF-to-MIDI · optical music recognition — Planning a tool to turn printed sheet music PDFs into playable MIDI and editable MusicXML, with a review-and-correction workflow. Starting with piano and single-instrument scores.
| Repository | What you’ll find |
|---|---|
zisheng |
Chinese-character diffusion pilot, saved model, and interactive sampler. |
rainpiao |
Website source built with HTML, CSS, and JavaScript. |
Rainnnee |
This README, the banner artwork, and scripts for the contribution calendar. |
🧩 Other technical work
- IMC Prosperity 4 — Python trading strategies, backtesting, and risk controls. Our four-person team reached Round 3 and placed in the top 16% of 18,803 teams.
- ASA DataFest at UCLA — Healthcare-data analysis with a five-person team, using R, ggplot2, and Spearman rank correlation to examine patient stress, social needs, and care patterns.
- LiDAR traffic monitoring — C++ breadth-first-search clustering to identify vehicles in 3D point clouds, form bounding boxes, and track movement.
Refreshed automatically about once an hour · Refresh status · Data & definitions


