Speech intelligibility prediction in noisy hearing aid scenarios, team of 7

Role : Deep Learning Engineer

• Designed and implemented a Whisper + Transformer-based model, achieved 7% reduction in prediction error and 9% increase in correlation compared to the baseline system

Ranked 5th among international participants

Currently conducting further advanced experiments to extend the research scope and preparing a paper for conference(IEEE TNNLS), Co-authored with Prof.Noori Kim, Purdue University

Implemented an augmented reality (AR) indoor and outdoor navigation system using the Unity platform to help university freshmen effectively learn and navigate the campus geography, team of 5

Role : Team Leader / Unity Developer / Technical Lead

• Led a development team, coordinating development schedules and internal team communication, while managing version control and code integration on GitHub

• Defined the project theme and designed the complete application flowchart

• Managed cross-Scene data persistence and information transfer within Unity, and oversaw overall project structure, integration, and asset management

• Implemented the Quiz and Reward scenes, utilizing dynamic user data for personalized content delivery

Application was used by approximately 60 freshmen and received an average
rating of 4.7/5

Developed a specialized website fingerprinting model for the Starlink network and identified security vulnerabilities through traffic analysis, team of 3

Role : Security Data Scientist / AI Engineer

• Performed Starlink network vulnerability analysis through detailed traffic analysis utilizing statistical methods, including HDBSCAN clustering

• Developed Starlink-specific fingerprinting model features by applying Quantile Transformer normalization to the inter packet delay (IPD) data, 8% increase in accuracy compared to the baseline system

Authored and published a full paper at Conference on Information Security and Cryptography (CISC); July 2025

Personal Project

developed a Discord chatbot leveraging AWS Lambda and Amazon SNS to provide students with convenient access to daily cafeteria menus on a 120-member Discord server

scraped menu data from the university website and responds to user commands in Discord via API integration, and sent menu information to users via email

Personal Project

Created a gallery website leveraging AWS EC2 (with Auto Scaling), AWS S3, AWS SDK(boto3) to feature and promote my photography shared on Instagram

Enabling image upload, display, and deletion

Developed a real-time computer vision model and an application designed for posture correction and monitoring in children , team of 4

Role: Vision ML Engineer 

• Developed the core vision model responsible for real-time posture analysis and detection.

• Implemented the application's foundational UI/UX, specifically designing and coding the initial Splash and Login screens.

Studied key robotics technologies including SLAM, YOLO, Reinforcement Learning (RL), and the Robot Operating System (ROS)

A hands-on project focused on mastering AI robot development, involving the construction of the integration of control systems (Jetson Nano/ROS) alongside the practical study of machine learning algorithms, particularly reinforcement learning, team of 4

Role : Robotics & AI Engineer

• Engineered an autonomous AI robot system by leveraging Jetson Nano and ROS for integrated navigation and object detection capabilities.

• Integrated the YOLOv5 deep learning model and performed the end-to-end assembly of the robotic hardware platform