a real-time augmented reality (AR) system developed as a HackMIT project to enable remote surgical experts to assist on-site medics via live vision, voice, and spatial tracking. The system uses XREAL glasses, a Raspberry Pi, an Auduicam, microphone input, and remote UI & data server components.

I am building a mentor–mentee recommendation system using a two-tower neural retrieval architecture, TF-IDF, and Sentence BERT. The data comes from my organization, Vietnamese Student Organization, where I am helping them reduce manual matching by reading mentor/mentee applications.

A financial data pipeline that fetches real-time OHLC (Open, High, Low, Close) data from Polygon.io API, processes it through Apache Kafka and Apache Spark for advanced anomaly detection, and provides actionable insights through a real-time dashboard. The entire system is containerized using Docker and orchestrated with Apache Airflow.

An interactive Streamlit dashboard that visualizes European call and put option prices under the Black-Scholes model. Users can tweak key inputs (spot price, strike, volatility, risk-free rate, time to maturity) and immediately see pricing heatmaps and P/L surfaces

The findings reveal distinct patterns in the types of residential buildings sold per borough and highlight significant variations in sale prices and transaction density, especially in Manhattan.

We scraped real-time data from Basketball-Reference.com, processed and cleaned it, and built both a statistical regression model and a user-friendly GUI to: 1. predict potential future MVPs based on their performance 2. Analyze which stats are most significant for MVP selection 3. Compare player-to-player performance and efficiency