Job Experience

Leading the Vision AI Insights Project: Applying Machine Learning and OCR Libraries to Solve Real-World Challenges with a Team of Three Members.

YOLOv10 Analysis:
- Analyzed YOLOv10’s innovative NMS-Free Training, highlighting improved real-time performance by eliminating the Non-Maximum Suppression step.
Effectiveness Evaluation:
- Demonstrated significant improvements in computational efficiency and model performance with YOLOv10’s Spatial-Channel Decoupled DownSampling and Rank-Guided Block Design for real-time object detection.
Cost-Effective OCR Solution:
- Integrated PyTesseract for image quality assessment and Amazon Textract for high-quality image processing, optimizing resource use and minimizing transaction costs.
Web Application Development:
- Developed a web application using HTML, CSS, JavaScript, and JavaFx, displaying OCR-extracted information, with backend processing and Docker deployment for scalability.

Led an OCR Project to detect container numbers, significantly reducing truck transit times at the gate.

Enhanced image detection accuracy by 20% by developing and refining a YOLOv5 AI model for improved feature recognition in container images.
- Improved image detection accuracy by 20% with a refined YOLOv5 AI model.
Reduced image processing time by 30% using OpenCV for efficient image parsing and pre-processing workflows.
- Cut image processing time by 30% using OpenCV.
Streamlined real-time data integration by automating JSON data transmission via POST API to an AWS server, decreasing data entry errors by 15%.
- Automated JSON data transmission to AWS, reducing data entry errors by 15%.
Managed over 10,000 daily data transactions on AWS.
- Handled over 10,000 daily data transactions on AWS.

Led the KPI Automatic Project, developed an executable program to automate data crawling tasks, significantly boosting operational efficiency.

Utilized Selenium to extract necessary information from web pages, achieving the desired result.
- Used Selenium to extract information from web pages successfully.
Automated data crawling tasks that typically took 20 minutes to complete manually, allowing for instant results with a click of a button and reducing task completion time by 80%.
- Automated data crawling, reducing task time by 80% and providing instant results.
Converted the application into an executable (EXE) file and distributed it to all employees, resulting in increased work efficiency and streamlined operations.
- Created and distributed an EXE application, boosting work efficiency and streamlining operations.

Scholarship

NAMU (New American University Scholar) Scholarship 2023 to 2025
- For top international students
FURI (Fulton Undergraduate Research initiative) Award
- For students who are eligible to participate the research program

Professional Experience