Wei-Chen Huang

Computer Architecture(A), Software Design Labs(A), Data Wrangling(A), Verification/Validation of Software(A-), Requirements Engineering(A), Javascript Programming(A), Programming Paradigms, Ethics of AI

Data Structure(A+), Computer Network(A+), Computer Architecture(A+), Software Studio(A+), Operating System(A+), Algorithm(A), Machine Learning(A), Computer System Administration(A), Database(A), Computer Graphic(A+), Parallel Programing(A-), Network Security(A+)

• Reduced multi-locale extension request task number from N to 1 per translation request by designing, developing, leading test and rollout stage of Bulk Extension Request workflow
• Eliminated 20+ admin hours/week through designing and building 1-click automated system as well as centralizingextension request management across different platforms and task types
• Resolved timestamp inaccuracies of 200M+ translation history rows, enabling the use of up-to-date data and ensuring accurate metadata on translation search portal
• Contributed 90+ pull requests in 3 months with near 100% test coverage by Hack, React, and Ent Framework
• Built new timestamp update logic and locale-level activity logs, improving traceability of time-sensitive translations
• Drove workflow adoption by orchestrating cross-functional team collaboration and delivering consolidation org-wide presentations (40+ people)

• Extending PEAK’s profiling framework to intercept and analyze CUDA kernel launch APIs, enabling fine-grained GPU performance evaluation in large-scale HPC applications

• Cut database traffic by 90%+ by implementing in-memory cache of product inventory in server-side
• Reduced ETL workflows cost and traffic by 50+% and file size by 26% by consolidating two pipelines (2M+ orders, 500K+ members) into a single Python pipeline on Google Cloud Run
• Built B2C interaction app to enable seamless customer engagement on Line App with Line Frontend Framework
• Conducted an early research and demoed of AI virtual customer service with text-to-video models in Vue.js
• Enhanced the user experience of a Generative AI solution using React, resulting in 1st place at an internal competition
• Created an automatic image carousel display component using Splide.JS that integrates AIQUA scenario-id data flow

• Boosted ad click-thru-rate 30+% by designing and implementing a ChatGPT-powered ad content creation UX in React.js
• Reduced initial network request time by 40% with asynchronous rendering multiple image recommendation ads
• Eliminated repetitive manual work by building bulk record update page to support 1000+ records
• Reduced survey misalignment by migrating legacy customer surveys onto the One Customer V oice platform
• Developed 10+ new features on the Recommendation Page using React.js, Backbone.js, TypeScript, C#, and CSS
• Increased test coverage by 15+% with 30+ Selenium/WebDriver based End-to-End and Unit tests in C# and JavaScript
• Conducted early research on analyzing customer feedback with ChatGPT for ad recommendation in Python

Legal Judgment Prediction - Porting the TopJudge Framework to Taiwanese Judicial Decision Data

• Proved transplanting possibility by porting existing Chinese prediction models to Taiwanese court documents
• Created a pipeline to process untamed Taiwanese legal judgment documents into structured data with Python

• Managed a 5-TA team, efficiently delegating duties to ensure classroom assistance of 120+ students
• Designed and evaluated 8+ assignments, weekly C/C++ online judge, exams, projects of Allegro5 and OOP
• Conducted weekly office hours, offering personalized guidance and providing a collaborative learning environment