README
I'm Nicole. Twelve-plus years designing at the seam where people meet machines: robots, then printers, then a superapp, now GIC's enterprise AI platform. Somewhere along the way I stopped being only a designer. I build what I design, and I run my life the way I run my projects: as a system under active development.
My core belief about the craft: design is decision-making. The most expensive problems were never visual. They were decided three meetings earlier, and the job is to be in that room.
My work so far
- Twelve-plus years across HP, Honestbee, Grab, FairPrice, and now Design VP at GIC, leading design for its enterprise AI platform. Details, with numbers, in /work.
- Trained in interaction science (M.S., Sungkyunkwan and Penn State). I started in human-robot interaction and never really left it. These days the robots are agents, and I'm asking a lot of the same questions.
How I prefer to work
- I do my best work where structure doesn't exist yet. My whole record is zero-to-one: new platforms, new functions, new ventures. Give me the room to build the structure and I'm at my best. Hand me a finished one and I get restless.
- The play I've repeated at every company since 2019: interview people one by one, then synthesize what they said with their own words as the evidence, then ask for two or three specific things. Influence works the way design works. Research comes first.
- I learned at Grab that you can't push a platform onto teams before they're ready to commit. So now I build the evidence that makes people pull.
- I schedule by energy. Coordination and management work get my peak hours; creative work gets the dips. I've run my calendar this way since 2020.
- Structure before build. Show me the shape of a thing before you make it. Once we agree on scope, run.
- I built a design function and managed people. I chose to go back to being the hands. The work should speak for itself.
- Most nights I scope a plan with an agent, let it run while I sleep, and review over coffee. Wall-clock time stopped being the limit on what one person ships.
How to work with me
- Disagree with me out loud. If you just agree with whatever I say, I trust you less.
- I can hear "this is wrong" without ceremony. Cushioning reads as noise.
- Critique with structure, not taste. "I don't like it" is not feedback.
- Corrected facts stay corrected. If I fix a name or a number once, I expect it fixed everywhere.
- Bring evidence. It beats the loudest voice in the room, including mine.
What I enjoy working on
- Systems that outlive the session: design tokens, agent harnesses. Building the machine that builds the thing.
- Craft in the details. That's where trust is built.
- Writing things down until they're true.
What I have no patience for
- Flattery shaped like agreement.
- Invented data. A made-up example is worse than an honest TODO.
- "Done" that nobody verified.
- Decisions made by the loudest or highest-paid voice in the room.
- Over-engineering. If the simple version works, the complex version is a hobby.
Quirks
- I dictate most instructions by voice, so my agents keep a decoder ring. "Karate skills" means Karpathy.
- I switch languages mid-thought. English is my working language and where I think and reflect. The hardest concepts need both English and Chinese side by side.
- Checkbox lists for everything, including feelings.
- I keep written playbooks for difficult conversations. Charisma, for me, is also a system.
The forest and the fences
In 2021 I wrote a line for myself: "I want to become a wild forest, not a manicured garden." Then I spent five years building fences: numbered project folders, an operating manual for my AI agents, a finance pipeline that never touches the cloud. If you only saw the fences, you'd call me a control person. You'd be missing why they're there. Nobody was coming to organise my life for me, so structure became how I make room for things to grow.
I want you to see the tension itself. It's like yin and yang, like entropy: everything dissolves into a mess, and maybe one value of being human is to create a beautiful experience out of that. There is no perfect balance. But there is an existence, and there is a presence. When we're struggling to strike the balance, the process itself is the truth.
The fences have a price. The perfectionism-driven ones have delayed a lot of opportunities to be true, and to be loud enough to put myself in front of others. This page went up before it felt finished. That's on purpose.
Frameworks I learned, then outgrew
I was a theory-first designer and a believer: Mori's uncanny valley, the Kano model, NPS at HP, Google's HEART and Design Sprint, IDEO. I started every problem from someone else's foundation, and back then I had no doubt about any of it. Meanwhile my own record contradicts me. I love to plan, I build organised backlogs, and a lot of those plans never got acted on. The work I'm proudest of lately started with no framework at all, just building while feeling like I don't know what I'm doing.
No idea is 100% anyone's; what makes an idea yours is that you paid for it. I paid for "testing is the most creative thing you can do for strategy" with three experiments at Grab. I paid for "set the kill bar before you see the data" with a strategy I had to bury. Those are mine. This page is partly me starting to name them.
The grader in my head
The best thing about me and my worst days come from the same place. I hear a false note right away: in a design, in a sentence, in a compliment. It's why employers kept handing me the zero-to-one problems. The trouble starts when that ear turns on my own work and becomes a grader, auditing drafts before they've had a chance to breathe. I've shipped platforms used by millions and still catch myself wondering whether an idea was really mine. I'd rather be seen accurately than admired.
Initiative runs on the same economics. It's the thing I get praised for most, and it's the thing that drains me fastest. Naming that is how I ration it.
Evidence over opinions
When a trading strategy promised 25 to 30 percent a year, I built a backtest, set the kill bar before seeing the data, and when the verdict came back kill, I wrote it in caps and moved on. When my own failures with AI kept repeating, I read 130 megabytes of my transcripts back and turned every repeat into a rule. I treat beliefs the same way: is this a fact or a story, and if it's a story, is it creating anything?
Two habits follow from this. I lock the measurement at kickoff, so success has a number waiting for it before anything gets built. And when a project gets killed anyway, I harvest it. The lessons and the artifacts still pay out long after the roadmap moved on.
Why this page is dated
My philosophy sits closer to Laozi and Naval Ravikant than to any design canon, and I keep this page dated because it's a trace as much as an introduction. The 2020 version of me collected other people's affirmations. The current version writes her own rules: consistency compounds; judgement is usually a gap in understanding; old stories get rewritten when they stop serving. I want to come back in ten years and see how the thinking moved.