At Mobispace Co., Ltd., I built and deployed a geospatial platform using React.js, Node.js, and MongoDB to plot coordinates and dynamic media for 1,200+ micro-spaces. I also represented the company at international conferences and sourced and briefed 10+ B2B corporations before handing qualified conversations to leadership.
In my telecom churn project, I built an end-to-end prediction pipeline on 72,000+ customer records. I found that missing contract values carried a strong churn signal, compared five classification models, and deployed a trilingual Streamlit app with the selected Random Forest model.
For human activity recognition, I engineered features from accelerometer data and compared SVM and Random Forest models across eight variants. I’m applying this hands-on data science and machine learning training alongside my software engineering background.
