Work Experience
TSMC (Taiwan Semiconductor Manufacturing Company)
Senior Software Engineer
Hsinchu, Taiwan | April 2023 - August 2025
Technologies: Python, FastAPI, Spring Boot, MariaDB, REST APIs, Shapely, Three.js
- Developed multithreaded, real-time parsers (10,000 logs/sec) to power exposure-aware models, increasing throughput by 100+ masks/year
- Built an end-to-end ML pipeline automating data ingestion, training, deployment, and monitoring, reducing deployment time by 70%
- Designed a scalable file system using state machines and message queues to orchestrate 50,000+ TB of production data across 30+ services
- Built 3DIC geometric verification engines to validate 100+ layout configurations, detecting die-gap, collision, and enclosure violations
- Collaborated with a 20+ member team to deploy a full-line 3DIC data pipeline, automating 100% of new 3DFabric business workflows
TSMC (Taiwan Semiconductor Manufacturing Company)
Software Engineer
Hsinchu, Taiwan | July 2021 - April 2023
Technologies: Python, Java, TensorFlow, TypeScript, PL/SQL, React.js, Vue.js, Docker
- Reduced mask-scrap rates by 60% (saving $3.5M/year) by developing a time-series ML anomaly detection system for mask writers
- Achieved 90% precision by developing a hybrid statistical/deep-learning model for distance and pattern-based detection for mask writers
- Re-architected seismic detection system from sequential file polling to a multithreaded socket-stream design, reducing latency by 88%
- Increased production efficiency 30% by building a full-line management platform that supports 200+ users across manufacturing sites
- Refactored codebase to Vue/React and migrated 50+ systems/services to Azure DevOps, reducing development cycles by 60%
Deloitte
Software Engineer Intern
Taipei, Taiwan | September 2020 - March 2021
Technologies: Python, PyTorch, Selenium, MySQL, React Native, Azure Cloud
- Improved BERT-based NLP model accuracy by 10% for risk advisory data via t-SNE analysis, correcting 1,000+ labels within 5,000+ articles
- Developed an automated Python/Selenium pipeline to scrape and preprocess datasets for NLP models, cutting cycle time by 90%
- Engineered a cross-platform (Web, iOS, Android) workflow system to automate review and approval requests, improving efficiency by 30%
