About Me
MS CS student at Duke specializing in ML/deep learning. Published on cat-breed classification and built a LoL win predictor and trading simulator; proficient with Python, PyTorch, YOLO, React, and FastAPI; UC Davis BS, GPA 3.9.
Education
- GPA: 3.9/4.0
- GRE: 339
Experiences
- Integrated Vue.js and Spring Boot for enhanced UX/UI and performance
- Boosted YOLOv3 image recognition accuracy through curated datasets
- Established Prometheus-based system monitoring with Grafana
- Graded assignments and provided feedback to improve problem-solving skills
- Grade assignments and provide prompt, constructive feedback
- Host weekly office hours for Q&A and debugging
- Lead pre-exam review sessions with practice materials
Projects
WOAA-trading
- full-stack trading simulation platform using React (TypeScript), FastAPI (Python), PostgreSQL, and Redis.
LOL-Win-Prediction
- Reached 98.9% clean accuracy (best variant) for DNN model while keeping predictions stable under perturbations
Chinese Handwritten Digit Classifier
- Achieved 95% accuracy on a 15,000-sample dataset using Logistic Regression and Neural Networks.
Cat Breed Classification
- A ML Classifier using YOLOv5 and VGG16, achieving 87% accuracy across five breeds
ResCash
- A decentralized bookkeeping app with React and Node.js.
BlackJack Brawl
- A strategic card game blending blackjack mechanics and unique card effects.
Publications
2024 International Conference on Modern Logistics and Supply Chain Management (MLSCM 2024), Oct 2024