Binary Segmentation for Cone Detection
Developed and trained a lightweight binary segmentation model using a simple encoder-decoder architecture to detect traffic cones in real-time for autonomous driving applications especially in the FSAE Driverless context. Achieved high accuracy and low latency, making it suitable for deployment on edge devices.

Optimized Translator Model for Jeju Dialect
Focused on developing an optimized embedding method for the Korean dialect Jejueo, leveraging jamos. Explored optimized translation by combining word identification with one-to-one mapping, aiming to facilitate the use and understanding of Jejueo and preserve the language.

Modular Mobile Robot with Computer Vision
Constructed a mobile robot capable of line tracing using computer vision (OpenCV). Hardware includes Jetson Nano, aluminum profile frame, and 9V 4A battery. Implemented PID control for motors and Kalman filter for sensor inputs. Currently integrating ROS2 and LiDAR for autonomous navigation.

VisioMed: Smart Voice-Guided Medicine Storage
Designed a smart storage system to aid visually impaired and elderly individuals. Integrates auditory reminders, NFC tags for medication ID, and computer vision to verify user actions. Powered by Raspberry Pi with magnetic and ultrasound sensors.
[View Code/Details]

Cost-Effective Liquid Handling Device for PCR
Developed an affordable liquid handling device for PCR using 3D printing and aluminum profiles. Features an H-bot system for XY movement and lead screw for Z-axis. Costing approximately $250, it offers a scalable open-source solution for underfunded labs.
[View Code/Details]

Web Application for BIPOC Community Wellness
Developed a prototype for TruDiary to support BIPOC women during pregnancy. Features include user authentication, location tracking via Google Maps API, and a messaging system. Built with React.js (Frontend) and MongoDB (Backend).

Sentiment Classification of Financial Comments
Compared Logistic Regression, LSTM, and BERT models for classifying stock-related comments. Rigorously evaluated using accuracy and F1-scores. The best model was integrated into a real-time system for market sentiment analysis.
[View Code/Details]

Solar-Powered Self-Driving Car
Designed a solar-powered car with infrared sensors for autonomous line-following without microcontrollers, focusing on analog circuit design.

Stock Market Volatility Analysis (COVID-19)
Analyzed the impact of the pandemic on major stocks using Python. Employed technical indicators (moving averages, Bollinger Bands) and predictive models to forecast price movements.
[View Code/Details]