Hi, I’m Nico. I’m a service and product designer in Vancouver.

I make complicated services flow, and build the tools they run on. One of them is live and self-hosted, teaching real families every week; the rest I hand over documented, so a team can run them without me.

selected work
↘ service design 2026

The Human platform.

From service model to daily operations, self-hosted for about $30 a month.

role
Founder; service design, build, operations
methods
Service blueprinting, role modelling, bilingual service design, AI review workflows
output
A live platform: scheduling, classrooms, lesson records, billing

My tutoring practice ran on six apps and a lot of memory. I designed the service model and built the platform that replaced them: one sign-on for four roles, Mandarin beside English on every family surface, and AI-drafted lesson notes that only a teacher can publish. Live with families since June 2026.

read case study ↗
↘ service design 2026

Day One.

Reintegration support at the moment of release from Canadian federal prison, redesigned after research broke the first concept.

role
Service design on the back end; data-flow mapping, regulatory audit, cost case
methods
Practitioner interviews, jobs to be done, archetype personas, data-flow mapping, a three-pass regulatory audit
output
Current-state and corrected data-flow maps, an AI field map with named refusals, a pain, fix and rule board, a tagged cost case

We spent six weeks designing a phone-based reintegration service. Then four interviews established that nobody is released with a phone and institutional IT is air gapped. The design moved to the parole officer, the only person present at both ends of a sentence, and a three-pass audit against the Privacy Act and sixteen commissioner’s directives rewrote the system behind it.

read case study ↗
the gate hestia the officer nostos the person
↘ service design 2026

An art studio, read as one lifecycle.

Six departments rebuilt over one shared record, with a self-hosted AI behind a human gate.

role
Service design, blueprinting, AI architecture
methods
Service blueprinting, stakeholder mapping, adversarial critique, AI systems design
output
To-be blueprint, AI architecture, per-role dashboards

A seven-person art studio ran on five workflow boards that barely connected. I read them as one student lifecycle and rebuilt it department by department. My first service-design project.

read case study ↗
before after one record · five states · loops back
↘ macOS app 2026

Chorus

One quiet window for the web apps you never close.

role
Product & interaction design, native build (solo)
methods
Interaction design, notification design, SwiftUI
output
Shipped macOS app: spaces, per-service notification control, sealed sessions

I gather the services you keep open all day into one native window, then spend the design on the hard part: keeping a dozen apps from all shouting at once. You govern the notifications, spaces match how your day splits, and two accounts of the same service stay sealed from each other.

read case study ↗ see on GitHub ↗
The Chorus GitHub repo shared in a UX design cohort's Slack, showing 22 stars and members signing up to test the app.
↘ web app 2025

Useful Words

A bilingual reference tool for language learners who get stuck mid-sentence.

role
UX/UI design, research, front-end
methods
Classroom observation, contextual interviews, prototyping, usability testing
output
Live bilingual web app, indexed by writing move, language-tagged for screen readers

My students kept switching between writing apps and dictionaries while drafting English sentences, losing their train of thought every time. Useful Words sits in one screen with examples, translations, and connector phrases ranked by how learners move through a sentence. Mobile-first because that is where the writing happens.

read case study ↗ try it live ↗
↘ web app 2024

Pocket Dystopia

A premise generator that gives fiction writers something specific to react to.

role
UX/UI design, build (solo)
methods
Prompt-structure iteration, design critique
output
Live web tool: five lockable slots, mood bias, shareable build codes

I rewrote the generator toward stranger, more specific seeds, and a design critique tightened the interface. It was never meant to generate a good story, just to give someone a useful place to start.

read case study ↗ try it live ↗
↘ macOS app 2025

WatchMeType

A typing simulator that produces exactly the revision history a school accepts as proof.

role
Concept, UI/UX design, build (solo)
methods
Speculative design, SwiftUI prototyping
output
Source-available macOS app, seven releases, 663 downloads

I built it because the conversation about AI in classrooms had become a checklist of what to ban. WatchMeType reproduces a human typing rhythm, varied pauses and corrections included. It can fool a teacher, easily. The harder question is what we are trying to detect, and whether teaching has to mean policing.

read case study ↗ view source ↗
↘ curriculum design 2017–2025

Eight years of curriculum design.

The unit plans I built had a lot in common with service blueprints.

role
Designer, facilitator (eight years)
methods
Instructional design, iterative material redesign, learner observation
output
Lesson plans, handouts, assessments, learner-centred materials

For eight years I watched where students got stuck, restructured the material, and tested whether the change worked. The vocabulary changes when I switch industries; the instincts carry over.

read case study ↗
↘ macOS app 2026

Quip

Making an invisible action tell the truth.

role
Product & interaction design, usability testing (solo)
methods
Interaction design, usability feedback, SwiftUI
output
Shipped macOS app, 17 releases, open source

A menu-bar app that finds a GIF and drops it on the clipboard in a click. A copy leaves nothing on screen, so the design had to prove it worked, and the best fixes came from watching real people use it.

read case study ↗ get Quip ↗
↘ campaign strategy pre-UX

Aoni Campaigns

Audience research that made the case for humour.

role
Audience research, campaign copy
methods
Post-level analytics, before-and-after comparison against the page's own baseline
output
Forty posts over five months; one format produced 92% of all likes

A Hong Kong condom brand entered the North American market on a hard sell, and the first six posts averaged seven likes. The format that replaced it asked for nothing and earned eleven times the response.

read case study ↗
experiments

Tools I built for my own practice.

notes read all ↗

I write these to work something out.