DANCE COMMUNITIES / EVENTS / CONNECTIONSswing
scene
radar.CONCEPT COVER · NOT A PRODUCT SCREENSHOT
Reliable event data, from unreliable sources.scene
radar.CONCEPT COVER · NOT A PRODUCT SCREENSHOT
DATA SCIENCE + QUALITY ENGINEERING
Hi, I’m
Ping-Lun
Chen.
An engineer who likes getting to the bottom of things. I turn manufacturing data into better inspection decisions—and build software that makes messy information useful.
Previously at Corning
M.S. Data Science, The University of Texas at Austin
01 / SELECTED WORK
Drag to explore · swipe sideways · Shift + wheel
LANGUAGE / LISTENING / LEARNINGHello.
Olá.CONCEPT COVER · ILLUSTRATIVE WAVEFORM
Feedback, down to individual sounds.Olá.CONCEPT COVER · ILLUSTRATIVE WAVEFORM
CONCEPT COVER · NO PRODUCTION DATA
Compare the rule before adding another check.PRODUCTION / IMAGE PROCESSINGA clearer
criterion.Repeatable defect classificationCONCEPT COVER · NOT MEASURED OUTPUT
From subjective judgment to production classification.criterion.Repeatable defect classificationCONCEPT COVER · NOT MEASURED OUTPUT
PROCESS / EXPERIMENTAL DESIGNMany factors.
Fewer suspects.Screening design of experimentsCONCEPT COVER · NOT MEASURED OUTPUT
Screen the causes, not just the finished product.Fewer suspects.Screening design of experimentsCONCEPT COVER · NOT MEASURED OUTPUT
INSPECTION / MEASUREMENT SYSTEMSMeasure the
measurement.Calibration / variation / consistencyCONCEPT COVER · NOT MEASURED OUTPUT
Separate measurement variation from the process.measurement.Calibration / variation / consistencyCONCEPT COVER · NOT MEASURED OUTPUT
UT AUSTIN / PREDICTIVE MODELING2,304
perspectives.Satellite features → ground-level PM2.5CONCEPT COVER · NOT MEASURED OUTPUT
2,304 image features, with time kept in view.perspectives.Satellite features → ground-level PM2.5CONCEPT COVER · NOT MEASURED OUTPUT
UT AUSTIN / RELATIONAL DATA13 tables.
One cohort.Performance / time / confoundingCONCEPT COVER · NOT MEASURED OUTPUT
Thirteen tables and a fairer comparison of progress.One cohort.Performance / time / confoundingCONCEPT COVER · NOT MEASURED OUTPUT
Open a case to read the full project.
01 / 11
Illustrative imagery, not workplace photographs or measured results.
03 / A LITTLE CONTEXT
How I got here.
My background starts on the factory floor, not with a perfectly clean dataset.
I spent more than six years in display glass manufacturing, working on defects, inspection systems, and customer quality. The work meant understanding both what happened in the process and why our checks had missed it.
I later completed an M.S. in Data Science at UT Austin. Now I bring that engineering background to statistical modeling, machine learning, and the software projects I build.
2017–2024Corning Display TechnologyProcess engineering → Quality engineering
2023–2026The University of Texas at AustinM.S. in Data Science · completed
BEYOND THE JOB TITLE
There’s a language
thread here, too.
Mandarin, English, Portuguese.
And a little German.
I spent a year studying Portuguese in São Paulo. Alongside the technical work, languages are another part of my background—and a fitting connection to a project like Pronunciation Analyzer.
04 / HOW I WORK
The practical side.
Experience from work
and personal projects.
Data & machine learning
Python, R, SQL · PyTorch, scikit-learn
Statistical modeling, simulation, DOE
Quality engineering
Root-cause analysis, 8D / CAPA, SPC
Inspection strategy, NPI, qualification
DevOps & deployment
Git / GitHub, CI/CD, Docker, Linux / VPS
k3s, Terraform, Helm
Personal-project use, with AI-assisted implementation.
Agentic software development
Claude Code, OpenAI Codex, GitHub Copilot
Agent-assisted implementation, testing, and debugging
Personal-project workflows, with review and validation.
05 / THE RÉSUMÉ VERSION