Making manufacturing processes measurable, predictable, and explainable — from microstructure to production line.
I am a manufacturing intelligence engineer. What that means in practice: I make production processes measurable, predictable, and explainable — and I have spent nine years working my way down through the layers where that explanation has to live.
I started in 2014 as a CNC programmer at Surya Moldtech, writing multi-axis toolpaths for precision tooling. Then injection mold design at Yeon Technology, where I learned that geometry is negotiable and material behaviour is not. Then four years at Tunas Makmur Jaya Abadi designing agro-processing machinery — rotary dryers, conveyors, palletizers, dust collectors — end to end, from requirements through commissioning on a plant floor.
Chasing process failures far enough eventually stops being a process question and becomes a materials question. That is what took me to the National Taiwan University of Science and Technology, where I completed an M.Sc. in Mechanical Engineering and developed an infiltrated stir casting route for magnesium hybrid metal composites, combining Taguchi-designed experiments with SEM/EDX and XRD characterisation. It was published in the Journal of Applied Science and Engineering. The lesson that stayed with me was not the result — it was that the result meant nothing until the microstructure explained it.
In Taiwan I also designed special-purpose machines and robot cells at Growin Automation, including an automated battery-cell leak-detection station. Today I am a Process Quality Engineer at CATL Karawang, applying the same loop at gigafactory scale: observe the process, explain it with physics, test the explanation with a designed experiment, and hold it with statistical control. It is the same question I was asking at the machine in 2014, asked in parts per million.
What I want next is to take that question further than industry alone can. Manufacturing generates far more data than explanation, and the methods I know best — designed experiments and statistical control — have structural limits. Building the physics into the model is where I think the answer is, and that is what I am looking for a doctoral group to work on.
A principle with no origin is a slogan. Each of these was learned somewhere specific.
A measured property gain without a mechanism is a coincidence you have not disproved yet. I do not report an improvement until something explains it.
Where from: Learned in the composites campaign, where the first process window looked promising and the micrographs said otherwise.
Inspection detects; control prevents. Every part you inspect is a part you already made, so the leverage is almost always upstream of where the defect was found.
Where from: Reinforced daily on high-volume battery lines, where the economics of detection are brutal.
If a format change or a material change is foreseeable, absorbing it in software costs less over a machine's life than absorbing it in tooling — even when it loses the quotation.
Where from: The robot-versus-indexer decision on the leak-detection cell.
Confidentiality, unmeasured results, and unstarted work are all easier to defend when they are labelled. An unlabelled gap is read as an overstatement.
Where from: Why this site marks proposed research as proposed and withheld figures as withheld.
National Taiwan University of Science and Technology
Research on magnesium hybrid metal composites: infiltrated stir casting, Taguchi-designed experiments, and SEM/EDX and XRD characterisation. Published in the Journal of Applied Science and Engineering.
Grouped by the question each set of tools answers rather than by software vendor.
Turning defect investigation into process control on high-volume battery production.
Taking a failure down to the material, and proving the microstructure explains the property.
Designing experiments so the answer is attributable rather than suggestive.
Nine years from toolpath to commissioned machine, across tooling, machinery, and automation.
Special-purpose machines and robot cells, specified from the measurement requirement upward.
This site is designed, built, and deployed by me — Next.js, TypeScript, and Tailwind, end to end.
Professional membership and community organisation, mostly from the Taiwan years.
Member
Professional membership in Taiwan's experimental mechanics research community.
Organizer
Coordinated the Indonesian student community at NTUST — events, arrivals support, and representation.
Committee Member
Supported the organisation of overseas elections for Indonesian citizens in Taipei.
Delegate
Cross-cultural leadership programme and international collaboration.
Member
Community building among international students at NTUST.
Get in touch
Engineering roles, research collaborations, or a manufacturing problem you are trying to explain — all welcome.