Projects

Work & Projects

Selected work across research, industry, and entrepreneurship.

Research2024
Published

Unraveled Graph Contrastive Learning

Novel graph contrastive learning framework addressing representation collapse in collaborative filtering. Co-authored paper published in Data Mining and Knowledge Discovery (DMKD), Springer, Vol. 38.

PyTorchGNNContrastive LearningRecommendation
Industry2023–Present
Deployed

HFT Execution Pipeline Optimization

Designed and deployed automated futures and options strategies at VICI Holdings. Identified hidden latency factors and validated structural changes that reduced end-to-end execution latency by 50%+.

C++Low-LatencyPythonSystems
Research2021–2022
Completed

Malicious Blockchain Entity Detection

NCHC-funded system for detecting malicious accounts and smart contracts on-chain. Engineered ML models achieving 89.1% F1-score and 88.4% accuracy. Includes a web crawler for large-scale data collection.

GNNXGBoostPythonBlockchain
Industry2022
Completed

Graph Recommendation for CTBC Bank

Graph-based recommendation framework using LightGCN and MLP to mitigate cold-start at CTBC Bank. Feature engineering across 20+ SQL tables on the Cloudera Data Platform.

LightGCNPySparkPythonSQL
Startup2020–2021
Incubated

Plasly — Warehouse Management Platform

Co-founded at NTHU Garage (9th cohort), Taiwan's official university incubator. Built and deployed the RESTful API backend for a warehouse management platform targeting plastic materials companies.

REST APIBackendPythonStartup
Industry2020
Completed

Predictive Maintenance at AU Optronics

Predictive maintenance models for production line downtime at AU Optronics. 92% F1-score; reduced revenue losses by ~US$300K/year. Sensor data collected via Raspberry Pi.

XGBoostRandom ForestRaspberry PiPython