Jet Xuen Lee (Sky)

Full Stack Software Engineer

• Led a team of 8 in implementing a cross-checking application for OCR-generated and handwritten notes using Java (Spring Boot) and Angular, resulting in a 20% increase in data accuracy.
• Developed an in-house image labelling solution, enhancing the Machine Learning department's labelling interface efficiency by 30%.
• Generated monthly analytical reports using Elixir Repertoire and SQL, leading the design and iterative improvement of frontend GUI templates, which improved user satisfaction by 25%.
• Achieved 95% code coverage by implementing rigorous backend Java testing with JUnit, ensuring code integrity and quality per SonarQube standards.
• Optimized API calls and architecture, reducing memory usage by 40% and accelerating batch processing jobs by 30%.

• Built and integrated Angular-based workflow screens with Spring Boot backend APIs, replacing legacy PEGA workflows to support end-to-end transaction submission, draft saving, SLA tracking, remarks handling, and role-based actions across multiple business modules.
• Developed a rule-based case assignment engine with round-robin distribution, Out-of-Office exclusion logic, conflict-of-interest checks, and workflow type routing, ensuring fair and compliant task allocation across user groups.
• Rebuilt batch job and database ingestion pipelines in Java, replacing legacy agents to handle 200+ reports and streamline automated case creation.
• Designed RESTful APIs with robust validation, sanitization, and role-based access control for case lifecycle management.

Software Developer

• Developed a web application using the MERN stack (MongoDB, Express, ReactJS, NodeJS), facilitating the data collection process with Amazon S3 Bucket and resulting in data throughput increased by 50%.
• Collaborated with analysts to enhance data quality by processing images with Adobe Illustrator, Photoshop, and Python libraries (OpenCV, Shapely), resulting in a 20% improvement in data accuracy.
• Developed spatial data analysis software to generate digital building floor plans from multi-angle photographs, reducing manual processing time by 40%.