Hi, I'm Oliver

Developing a Mixed-Integer Programming (MIP) model for a multi-station production system to optimize scheduling across various material categories and work centers.

• Developed a MIP model for a multi-station production system to optimize scheduling across various material categories and work centers.
• Formulated complex constraints including dynamic inventory balance, BOM consumption, and station capacity limits with overtime considerations.
• Implemented an objective function to minimize total costs (delivery penalties + overtime labor) using Python and Gurobi optimizer.
• Modeled sophisticated logical relationships such as setup time triggers and warehouse volume capacities using Big-M formulations.
• (Expected Dec. 2026) Deploying a scalable decision-support tool balancing inventory holding costs against customer service levels.

Collaborated with an international team to solve YouBike supply-demand imbalances across the NTU campus by optimizing dynamic reallocation via worker scheduling.

• Collaborated with an international team to solve YouBike supply-demand imbalances across NTU campus.
• Engineered a data pipeline using Python to process raw trip logs, extracting hourly demand and utilizing KMeans clustering to analyze usage patterns.
• Formulated a MIP and developed scalable O(jdp) greedy heuristic algorithms with Python and Gurobi to minimize unmet demand and worker costs.
• Achieved near-optimal reallocation schedules — heuristic outperformed naive baselines by significantly reducing unmet demand with ~0.2s runtime on 48,000+ records.

Conducted mixed-methods research to investigate how cognitive decision-making styles (Maximizers vs. Satisficers) influence user interactions with navigation tools.

• Conducted mixed-methods research (Online Survey n=84, Email Interviews n=5) on cognitive decision-making styles in navigation tools.
• Performed data wrangling and statistical analysis using single and multiple linear regression models to evaluate behavioral traits.
• Identified statistically significant correlation (p < 0.05): Maximizers evaluate multiple routes more frequently.
• Concluded that highly structured digital recommendation systems standardize user behavior, reducing cognitive load and choice paralysis.

A clean, mobile-friendly Bible reader supporting Traditional Chinese (CUV) and English (KJV), built as a Progressive Web App installable on iOS and Android.

• Built a full-featured Bible reader supporting Traditional Chinese (CUV) and English (KJV) translations with complete 66-book navigation and chapter switching.
• Implemented font scaling and six colour themes for a comfortable and personalised reading experience.
• Added keyword search with jump-to-chapter functionality and auto-saves reading progress via localStorage.
• Deployed as a PWA — users can install it on iPhone or Android as a native-like app directly from the browser.