Reading the system before drawing the screen - A habit from my chip design years

TL;DR

Applied Electronics at Politehnica Bucharest, then four years at Microchip. I went from intern to Digital IC Design Engineer on the PIC16 and PIC18 families.

Microchip is one of the largest microcontroller makers in the world. Its parts sit inside washing machines, cars, and medical devices.

This is not a design project. It is where the way I work came from.

What the job was

I joined as an intern straight out of Politehnica Bucharest, Applied Electronics. I covered the full cycle from writing the hardware code to the final checks before manufacturing.

Design. Writing in Verilog, where every line ends up as physical transistors on a real part.

Verification. Building test cases in Assembly and running automated suites that recheck everything after every change. Find the flaw before the chip ships.

Post-gates checks. Reading Synopsys reports to confirm the design still works once it becomes actual hardware. The last gate before manufacturing.

The habit the job forced on me

In chip design you do not start by building. You start by reading.

I read the architecture, the constraints, and the concept, then looked for existing chips that solved the same requirement. I drew the block components on paper before writing a line of code.

Iteration on silicon takes months, so nobody gets a second attempt cheaply. You get it right the first time or you wait a quarter to find out you did not.

The same habit, in design

Reading first. Before I design a feature I read the design library, the live product, and the technical limits. I find out what can be built, then design inside it.

Breaking things before users can. Verification taught me to test my own work until it fails. I push every built feature against the design, next to QA and on my own.

Speaking engineer natively. Four years inside an engineering team, in their tools and their reviews. When I scope feasibility with developers now, I am not translating anything.

Where it paid off at Archbee

The zoom bug. Comments broke when the interface zoomed from x1 to x1.15. I read the code with the developer, then moved the range to x0.9 through x1. Same control, and the bug had nowhere left to happen.

The fake fades. Archbee faded content with white rectangles laid over it, so those areas stopped responding to clicks and people never realised they could scroll. I taught the developers to use real masks instead.

The constant one. When engineering pushes back I do not argue with the constraint. I find the concern behind it and design so the constraint stops being a problem.

The outcome

The chips shipped. PIC16 and PIC18 parts I worked on are inside products still running today.

Nothing I worked on ever triggered a recall. I caused a few in the other direction, by finding flaws in older chips that nobody had documented.

Why I left

New chips stopped being designed. Only small updates, and after each one the full verification cycle again.

My role drifted from designing circuits, the part I loved, into verifying the same things repeatedly.

I need work that feeds my brain, and repetition does not.

Why design

I am fascinated by how people build something new, and I studied the human decisions behind how things get made. I worked as a graphic designer in my spare time.

Graphic design had one flaw I could not live with. Everyone has an opinion and every opinion weighs the same.

UX is different. Every change has to be justified by intention, theory, and a real problem it solves, and an engineer can live there.

What this means for your team

Verilog became Figma. Gate-level reports became the preview environment, the last check before something reaches production.

I read the system before I draw the screen, so what I hand over already fits what your team can build. Nobody has to send it back.

I did not leave engineering behind. I brought it with me.