Ziyang (Rocky) Chen

Helped test a Hong Kong MTR control system, checking that equipment signals, backend services and operator screens showed consistent results.

• Checked 20,000+ monitoring and control points. Supported validation across field devices, servers, edge gateways and operator interfaces.
• Built automated tests with Python and Selenium. Turned I/O lists into repeatable test cases, so checks could run consistently at a large scale.
• Made complex web screens testable. Wrote JavaScript to reach controls inside nested Shadow DOM components that standard browser tests could not easily access.
• Made verification faster and easier to review. Automated checks and screenshot collection, reducing the verification cycle from 2–3 months to under 10 days while keeping traceable evidence.
• Investigated 10+ failures across system layers. Followed logs, service status and signal paths through Cloud Foundry services, Docker edge gateways and industrial controllers (PLCs) to find the cause.
• Supported system, factory acceptance and regression testing. Checked results against I/O mappings, recorded defects and test evidence, and tracked fixes and resolution status in JIRA.

Built an internal reporting dashboard that helped non-technical colleagues explore company and subsidiary data without requesting a separate report each time.

• Organized 2,000+ company records into a usable dashboard. Let users explore organizational levels, countries and cities through interactive filters.
• Developed the backend with NestJS, TypeORM and MySQL. Created shared APIs for filtering, grouping and summarizing company data.
• Modeled parent–subsidiary relationships. Used a self-referencing database entity and dynamic SQL queries to support three hierarchy tiers and six optional filters.
• Made repeated queries much faster with Redis. Cached frequently requested results, reducing targeted complex queries from several seconds to under 100 milliseconds.
• Protected 10+ REST APIs. Added JWT sign-in checks, validated incoming data and masked sensitive fields before returning responses.
• Developed five analytics and reporting pages. Used Next.js, Material UI and reusable charts to make the results easier to understand and explore.