· Analyzed 50k+ local retail transactions using Python (pandas, scikit-learn) and time-series forecasting models (ARIMA and Prophet), producing weekly demand forecasts that improved small-vendor inventory turnover by 8%

· Built customer-segmentation pipelines with K-Means and DBSCAN to identify underserved consumer segments, personalized insights by field teams and boosted sales by 12% for low-income vendors versus baseline periods

· Developed and deployed an interactive Tableau dashboard that scheduled extracts visualized sales, demand, and stock metrics, enabling non-technical shop owners to make data-driven pricing and purchasing decisions

· Conducted research in the Computational Combinatorial Optimization Lab on integrating mathematical modeling and AI-driven optimization for large-scale scheduling systems

· Integrated LLMs with operations research to build adaptive scheduling pipelines for constraint extraction, job classification algorithms, and reduced planning time by 20%

· Presented the research project at the 4th NUBS Tri-Campus Conference as the only undergraduate student

· Conducted market-trend and consumer-behavior analysis for Colgate, Always and Freepoint using Excel/SPSS and Python, evaluating product performance across SKUs

· Processed and visualized KPI dashboard for P&G in Python (pandas, plotly) to Tableau, improving reporting efficiency by 50%

· Scraped 2k+ live-streaming records and built a gradient-boosting classifier (F1: 0.86) to identify potential market growth opportunities

· Led early-demand interviews with 15 stakeholders, analyzed e-commerce trends and explored new retail models with log-log regression to size opportunities

· Built a 3-year EV market dataset with Python/SQL ETL (BeautifulSoup, OCR) to collect data on market size and policy data

· Quantified revenue/margin (around 16%) by assessing China’s new energy automobile industry under carbon neutral policy via scenario models