Huanxuan Li
Engineer. Researcher. Builder. Leader.
Huanxuan, also known as Shawn, is a junior at Thomas Jefferson High School for Science & Technology. Building at the intersection of AI, biomedicine, and robotics.
0+
Users
SkinAI + CAPA combined
0+
Students Reached
Growing Up with Robotics
0+
Service Hours
Community impact
Selected Work
Projects &
Research
Six projects across ML research, computer vision, nonprofits, and competitive engineering.
CAPA
Computational Architecture for Predicting Alloimmunity
An open-source ML framework using ESM-2 protein language model embeddings and deep competing-risks survival analysis (DeepHit) to predict GvHD, relapse, and transplant-related mortality in pediatric HSCT. Features a cross-attention interaction network over donor-recipient HLA pairs.
Python · PyTorch · FastAPI · Next.js
Competing-Risks Survival Analysis
Pediatric HSCT Outcomes Research
Solo-authored research applying competing-risks survival analysis and ML to predict acute GvHD in 187 pediatric bone marrow transplant patients. Severe aGvHD risk plateaus by day 100 (CIF=0.214). LASSO-selected Cox model achieved omnibus significance (p=0.002).
R · Python · lifelines · LASSO Cox
SkinAI
Computer Vision Skin Analysis Platform
AI-powered skin analysis platform with thousands of users and a pending patent. Uses computer vision for dermatological screening. Built from the ground up as a solo founder.
Computer Vision · Python · React
Growing Up with Robotics
International 501(c)(3) Nonprofit
International nonprofit I founded. Robotics and STEM education for underserved communities. 7,000+ students reached across multiple countries. Executive Director.
501(c)(3) · 300+ service hours · 7,000+ students
FTC Robotics — Team 14607
Robot Uprising · Team Captain
Team captain for Robot Uprising. Competed at FTC Chesapeake Championship. Led mechanical design and autonomous programming. Drove strategy and systems integration.
Java · CAD · FTC SDK
About
The story
behind the work
Hey! I'm a junior at TJHSST in Northern Virginia. I was born in 新丰 and immigrated from China in 2019. After arriving in the U.S., I taught myself English, started going to church and found a close group of friends. A the age of 13, I received a cord blood transplant at Children's National Hospital to treat beta-thalassemia, a genetic disorder that shaped my life at a very young age. Since then, transplant immunology and GvHD research haven't just been academic interests to me. They've been deeply personal.
I've conducted research at Stanford through iGEM, at George Mason University through ASSIP, and about to begin my senior year long research at Children's National Hospital. CAPA is my effort to turn that lived experience into something that could genuinely help the next child in that situation, while also contributing to the broader research that improves outcomes and advances the field.
“But they who wait for the Lord shall renew their strength; they shall mount up with wings like eagles; they shall run and not be weary; they shall walk and not faint.”
— Isaiah 40:31
Beyond the Lab
Competitive Speedcuber
Peaked Top 10 in states, Top 200 in country
Varsity Lacrosse
2-way midfielder
Professional Pianist
ABRSM certified · music instructor
Youth Volunteer
AYM 501(c)(3) · camps, workshops & outreach
Research
Awards & Honors
School Activities
Coursework
AP
Advanced
Publications & Research
Research &
Writing
Competing-Risks Survival Analysis for Pediatric HSCT
Applying competing-risks survival analysis and ML to predict acute GvHD in 187 pediatric bone marrow transplant patients. Severe aGvHD risk plateaus by day 100 (CIF=0.214). LASSO-selected Cox model achieved omnibus significance (p=0.002).
View on GitHub ↗CAPA: Computational Architecture for Predicting Alloimmunity
Full framework paper describing the ESM-2 + cross-attention + DeepHit architecture for structure-aware HLA mismatch scoring and competing-risks survival prediction in allogeneic HSCT.
View on GitHub ↗Data-Driven Prediction of Metal-Organic Dissolution and Adhesion Using Machine Learning
Co-authored published paper applying Random Forest and SVM models trained on 71 experimentally validated interactions to predict metal ion dissolution and adhesion using molecular descriptors and environmental variables. Published in GMU's Department of Civil, Environmental, and Infrastructure Engineering.
View Paper ↗Stanford iGEM — Immunocloaking
Synthetic biology research on immunocloaking strategies. Contributed to experimental design and computational analysis as part of Stanford's iGEM team.
Skills & Tools
Technical
Expertise
Languages
AI / ML
Biomedical
Engineering
Music
Business & Research
Contact
Let's talk
about ideas.
Research collaborations, internship inquiries, or just want to connect — reach out.
Send an email ↗