About me
I am a software engineer with 3+ years of experience, focused on Python backend development and system architecture, with frontend experience in JavaScript and React — comfortable working from the database and API all the way to the user interface. I graduated from the Department of Computer Science and Engineering at Yuan Ze University.
I am curious about new technologies, enjoy challenging myself, and like applying what I learn to real-world projects.
Work experience
Current
On-site Engineer
緯致科技股份有限公司On-site at Wiwynn (緯穎科技服務股份有限公司)
Building and improving internal systems and workflows for the company's hardware engineers.
A company-wide platform for checking equal-length PCB traces. It replaced manual Excel reviews with an online workflow, cutting review time from about a week to one day.
- EQL (equal length): signals in a high-speed bus must arrive at the same time with matching electrical characteristics, so their traces need near-identical lengths; the system manages these trace groups
- Replaced manual Excel-based checks with an online web workflow now used across the company
- Cut review time from about a week to one day, with auto-generated summary reports to verify trace design quality
- Used mainly by EE (electrical), PWR (power), and Layout (PCB routing) engineers
Automated schematic conversion: turnaround dropped from about half a day of manual work to one hour, with automatic failure alerts and cloud upload.
- DSN is the schematic file format of OrCAD (circuit design software); the system combines OrCAD built-in commands with AutoIt (a Windows automation scripting tool) to convert schematics automatically
- Reduced conversion time from about half a day of manual work to about one hour
- Uses AutoIt to watch for OrCAD errors; failures caused by OrCAD's own instability happen about once a week, and the failure log is emailed to the engineer automatically with hints on what to fix
- Uploads converted files to the cloud for engineers to download for downstream work
Mid-level Engineer
和瑩電腦股份有限公司
Worked on the National Fire Agency (Ministry of the Interior) AI Smart Search and Rescue Dispatch System project, and handled daily data center operations.
The aerial rescue request and approval platform used by every city and county in Taiwan, alongside central search and rescue agencies.
- In use by local agencies in every city and county in Taiwan, as well as central search and rescue agencies
- Flexible application and approval workflows for local agencies, based on BPM (business process management) logic
- GIS (geographic information system) map integration to show case locations, with real-time notifications for new requests
- Rewrote Django Middleware to integrate with the SSO (single sign-on) authentication system
- Wrote JMeter load-test scripts and passed the performance acceptance requirements
- Delivered six training sessions across North, Central, and South Taiwan, and wrote the training materials
One place to manage API integrations, usage, and access across 8 systems.
- Connects 8 systems, routing requests between them through an Nginx reverse proxy
- Monitors API usage per system and can disable specific APIs when needed
- Uses Celery (a Python background task scheduler) to pull Nginx logs on a schedule and analyze API usage
Real-time radio control at disaster scenes, with live speech-to-text transcripts.
- Uses WebSocket (a real-time two-way browser–server connection) to show the speaker, time, channel, and content live
- A second WebSocket uploads content to the central system in real time
- Integrated Google Speech-to-Text to transcribe radio audio live
Central hardware registration with live monitoring dashboards.
- Built central hardware registration on Django Admin (Django's built-in admin site)
- Integrated Prometheus (metrics collection) and Grafana (dashboards) to show device status in real time
Shrimp Fry Real-time Counting
Fine-tuned a YOLOv7 object detection model to count shrimp fry from camera feeds, at about 89% accuracy.
Test Engineer Intern
安智聯科技有限公司
Worked on factory systems: SPC (statistical process control), AOI (automated optical inspection), and RPA (robotic process automation).
- Built an automated testing framework with pytest and Selenium (browser automation)
- Developed backend APIs with Flask
Graduate Research Assistant
Yuan Ze University
Locates and segments the liver from CT scans, predicts likely lesion locations, and shows the result in 3D in the browser.
Outstanding Award, internship project contest
- A CNN (convolutional neural network) reads patient CT images to detect the liver's position
- A 2D U-Net segmentation model isolates the liver from the rest of the body
- Three.js (a web 3D graphics library) renders the liver model in the browser
An online testing system that recommends questions based on each user's results.
- Generates questions using two recommendation approaches: content-based (CB) and collaborative filtering (CF)
- Classifies questions automatically with an Academia Sinica model
- Recommends questions to users based on their test results