2023–2024
SGS
Third-party laboratory
Observe failure
Evidence is manufactured, not found. I learned where a fixture can lie to you, and what a defensible conclusion requires.
Hardware engineer at Amazon · Product integrity & NPI
I study how products
earn the right to exist.
A physical product has to survive reality, earn confidence, and make commercial sense. Someone has to decide when all three are true.
I have made that decision from three seats on the same supply chain: a test lab, a system company, and a global brand.
- LinkedIn impressions
- 48,707
- Sep 2025–Sep 2026 · 17,884 members reached · +38% year over year
- LinkedIn profile ↗
0.1
Three questions I keep answering
Most engineering writing explains methods. These pages explain decisions: harder to learn, and rarely written down.
-
How a product gets designed
How do partial truths become one coherent product?
No one designs the whole product. Every function designs a different version of the product. The customer only gets one.
-
How a product earns confidence
How does NPI turn failure into confidence?
A working design is only the beginning. NPI has to turn failure into knowledge, correction into evidence, and evidence into confidence at scale.
-
How a product creates value
How does a customer problem become a product that keeps solving it?
A request is only the beginning. Product development turns the problem behind it into something real, then keeps learning from use.
0.2
Tested by one company.
Tested by one company.
Built by another.
Decided by a third.
The same product looks different from each seat on a supply chain. I have sat in all three, and the disagreements between them are where the real engineering happens.
2024
Delta Electronics
System company
Understand design
Simulation and automation showed me how early design choices set the cost of every later iteration. Much of engineering is removing friction between tools.
2025–2026
Amazon
Global consumer brand
Decide readiness
Across EVT, DVT, PVT, HVT and mass production, the question stops being "did it pass" and becomes "do we ship". Evidence, statistics and business context arrive at the same table.
0.3
Working under confidentiality
The product can be confidential. The thinking does not have to be.
I cannot publish the devices, datasets or conclusions I work with. For a while that seemed like a reason to write nothing. Then it became obvious that the confidential part was never the interesting part.
What transfers is the reasoning: how to frame uncertainty, how to design evidence worth trusting, how to recognise a failure mode that is hiding, and how to turn technical context into a recommendation somebody can act on. None of that belongs to any one employer.
This site argues about method. The programmes stay out of it.
0.4
I like explaining this out loud
Engineering also means explaining a decision to people who did not run the test.
Techie Talk
Shared what hardware engineering looks like inside a global consumer brand: from design to mass production, the decisions that get hard, and how AI can support the workflow without replacing the engineer.
SPIE Smart Structures + NDE 2026
Presented my master’s research, first author: a 9 × 6 × 0.5 mm ultrasonic piezoelectric plate motor with vision-based closed-loop control. Feedback cut orientation drift from 10.2° to 2.6°, reduced straight-line error by 82%, and let the motor carry nearly five times its own weight.