Data Science + Economics @ UW Seattle · graduating June 2027 Building at the intersection of ML, systems, and things that actually ship.
- IsoCourt — AI-native badminton coach using CNN-LSTM + Gemini. Real-time stroke recognition, pose estimation, and coaching feedback from raw video.
- Eat Together — Leading a team of 18 to build a community platform connecting people over meals (550+ users, React Native + GCP)
- SWECC Labs — Mentoring 35+ students in open-source development across 38+ repos
| Role | Where | Impact |
|---|---|---|
| Data Science Intern | FlyDubai | Flight delay models on 600k+ rows (R²≈0.80), 30% sentiment accuracy boost via prompt engineering on Amazon Bedrock |
| Director | Eat Together | Shipped recommendation systems, reduced code review time 20%, scaled to 550+ users |
| Software Engineer | SWECC | Led Labs division, improved job search workflows for 2000+ engineers |
Languages: Python · Java · JavaScript · SQL · R · HTML/CSS ML/DS: Scikit-learn · TensorFlow · XGBoost · LightGBM · Pandas · NumPy · GeoPandas Web: React · Node.js · Svelte · React Native Cloud/Tools: AWS (Lambda, S3, EC2, Amplify) · GCP · Firebase · Docker · Azure DevOps
IsoCourt — CNN-LSTM + Gemini coaching platform for badminton. ResNet-50 feature extraction, MediaPipe pose estimation, sliding window pipeline. repo · live
DataMed — Healthcare cost modeling on 1.2M records. Optuna hyperparameter tuning. 3rd place — DubsTech Datathon (Best ML Model). repo
Macro Scope — IMF macroeconomic analysis on 400k+ records. XGBoost + grid search CV reduced RMSE by 20%. repo
NCAA Basketball Analysis — 20 years of data (300k+ rows), found 25% correlation between free throws and tournament wins. repo
_Open to Data Science / Data Engineering / SWE internships and new grad ro