Engineering Around
Real Business Problems
I'm Cai Yulin, an FDE engineer — on Chinese factory floors, people address me as "蔡工" (Engineer Cai). The AI implementation services on this site are provided by Guangdong Qingxian Technology Co., Ltd., delivered by me.
NOTE / On my name
Why "蔡工"? In Chinese manufacturing, "surname + 工" is the standard respectful way to address an engineer — short for 工程师 (engineer). Over twenty years, from R&D to the production line to customer sites, that is what colleagues and clients call me.
It is easier to remember than a full name, and more grounded than a job title: being called "蔡工" means the work is expected to be solved on site, not just in a slide deck.
With a mechatronics background, I have worked in the health equipment industry since 2005 — across R&D, production, and sales, more than 20 years on manufacturing floors. I have seen the real constraints of production lines and I understand what sales teams lose time on every day.
Today I apply that on-the-ground experience to AI engineering. My focus is not showcasing model capabilities in isolation, but bringing business processes, data, integrations, human review, and acceptance criteria together. I start with a specific workflow, clarify what information is available, who owns decisions, and what counts as success — then build a testable solution.
Critical judgment stays with people, and unverified assumptions are labeled as such.
Background
- Mechatronics (college diploma)
- 2005–present: Health equipment industry — R&D → production → sales, full chain, 20+ years
- Now: Independent FDE consultant focused on AI adoption in manufacturing and export
Why a manufacturing veteran doing FDE?
I know the floor
Over 20 years I drew parts in R&D, chased mass production on the line, and followed customers in sales. I know what a "simple" requirement really means on a shop floor — which decides whether an AI solution lands or dies in a slide deck.
I know engineering
Most AI projects fail not because the model is weak, but because process, data, permissions, and failure handling were never designed. I treat AI as engineering: define the problem and acceptance first, keep humans approving critical actions.
Chose the hardest path — and the right one
There is a saying in the industry: being an FDE is an invisible entrepreneurship bootcamp — product risk, client risk, and engineering risk, all carried by one person. The only difference: at a large company the employer absorbs the financial risk; on your own, you carry it yourself. I apply that training directly in my own lean model — no big team, no platform lock-in, staged delivery with acceptance on results, passing the saved overhead back to clients.
Let's talk about your process
A 30-minute conversation tells whether I can help.