Taipei, Taipei City, Taiwan
Senior Data Scientist with 5+ years of experience delivering production-grade ML solutions across large-scale, complex operational systems, specialising in Advanced Machine Learning (Unsupervised, Supervised) and Enterprise Data Solutions (Statistical Data Analysis, Interactive Data Visualizations). Skilled in Exploratory Data Analysis, Predictive Analytics, Time Series Forecasting with ML, and Cloud-Based Workflows, with a strong focus on developing data-driven insights to optimize aviation operations (Passenger Boarding Modeling, Risk Model of Cabin Crew Standby Planning, Probabilistic Model of Flight Crew Trip Assignment). Background in Business Information Management (Data Science) and Econometrics, with experience working across UK, Netherlands, and Taiwan. Certified in Harvard's CS50 Computer Science for Artificial Intelligence, and MIT IDSS's Data Science & Machine Learning. Motivated by impactful problems, collaborative teams, and applied ML that drives real-world decisions. Currently seeking my next high-impact Data Science opportunity in London or in Taipei.
Cabin Crew | Standby Demand & Operational Risk - Developed probabilistic models to support cabin crew standby planning, combining demand distributions and supply levels to quantify operational risk - Designed risk metrics (coverage risk, expected shortfall) and linked them to downstream business outcomes such as delays, cancellations, and financial costs - Extended an existing production architecture with a new risk-modelling pipeline, implementing it as a Dagster asset with clear upstream and downstream dependencies Flight Crew | Assignment Optimization & Trip Coverage - Built Time-Series ML models to forecast the probability of trip assignment drops and pickups from D-14 to D-0 (Day-of-Ops), providing decision signals to an optimisation engine - Applied a Direct Forecasting strategy, with step-specific models and dynamically updated feature sets - Engineered features from exogenous trip attributes and windowed roster history to reflect the latest known operational state at each forecast step
Passenger | Dynamic Gate Agent Assignment - Built predictive models to improve boarding punctuality through boarding dynamics estimations (growth rate and inflection point) from flight attributes and passenger profiles - Applied unsupervised learning methods (Principal Component Analysis, KMeans) to cluster boarding curves, and used model inspection techniques (Partial Dependence Plots, Accumulated Local Effects) to identify key drivers and actionable thresholds for operational interventions - Delivered a modular boarding analytics framework reusable across multiple operational use cases Passenger | Hand Luggage Prediction - Developed predictive model to estimate hand-luggage demand using booking data, aircraft characteristics, and route information Service Recovery | Disruption Management Analytics - Designed dashboards to monitor flight disruptions and customer rebooking performance, improving visibility into recovery speed and service bottlenecks
- Created 30+ Data Tables, combining from multiple data sources and across various operational units - Developed 5+ Enterprise Dashboards, creating an unified entity view to serve each BU's need - Organized Task Schedulers, setting workflow dependencies to pull data updates
- Codeveloped data generation & data validation script to produce reliable weekly reports - Implemented parsing rules to capture specific texts - Conducted ad hoc analysis to support various company requests