Product Designer

I design with AI in the loop, not bolted on after.

Irwin Yi — a Melbourne-based product designer who pairs UX craft with AI tools at every stage: synthesising discovery, generating personas, and prototyping working software in hours instead of weeks.

Portrait of Irwin Yi
Irwin YiMelbourne, AU
Tools in active rotation
Figma Figma Make Microsoft Copilot Claude FigJam / Miro
01 / About

A designer who ships, not just wireframes.

Irwin is a product designer currently working at Publicis Sapient, a global digital consultancy, where he's focused on turning ambiguous business and customer problems into designs that hold up in front of executives and developers alike. Across engagements he has worked on public sector, energy, insurance, and urban planning products — often stepping into projects with no prior context and a short runway to deliver, which has sharpened his ability to get up to speed fast in unfamiliar domains.

Over the past year, AI has become a core part of how he works: using AI for discovery synthesis, persona generation, and rapid interactive prototyping that would traditionally take a team weeks to produce. Several of these AI-assisted prototypes, originally built as personal experiments, have gone on to become client-facing deliverables.

He holds a Bachelor of Industrial Design (Honours, First Class) and previously lectured in Industrial Design at RMIT University. He holds an Australian Department of Defence Baseline Security Clearance.

  • Based inMelbourne, Australia
  • Current roleProduct Designer, Publicis Sapient
  • EducationBachelor of Industrial Design (Hons, First Class)
  • ClearanceAustralian Dept. of Defence — Baseline
  • CitizenshipAustralian citizenship
    Portuguese citizenship — full EU working rights
  • FormerlyLecturer, Industrial Design — RMIT University
UX Design Product Design Design Systems Concepting Wireframing Prototyping User Testing
02 / Selected work

Three engagements, each pushed further with AI.

Field operations · Public sector energy

Redesigning a field portal, with AI compressing discovery into days

Client: a government-owned electricity network operator. Brief: redesign the legacy web mapping application field crews use to plan and run operational activities across the network.

Engagement2 phases, ~8 weeks total
RoleLead designer (stepped up to Senior Manager scope)
Output170+ dev-ready screens
Tools used
Figma Microsoft Copilot Fireflies.ai

What happened

  • First engagement: an individually-led two-week sprint delivering a heuristic review and a 32-page best-practice report, operating at a Senior Manager level out of project necessity.
  • The client returned six months later for a six-week follow-on to design the full field portal map — using that original report as the north star.
  • Co-facilitated a two-day in-person discovery workshop and field shadowing session with operational crews.
  • Used Fireflies.ai to record and synthesise discovery sessions, generating faster, more accurate, and less biased insights than manual note-taking.
  • Built six personas using Copilot, ingesting raw discovery data to rapidly anchor the design direction.
  • Co-designed 170+ high-fidelity, dev-ready screens for handover.

Skills exercised

Discovery facilitation Field research Heuristic review AI-assisted synthesis Persona generation w/ AI High-fidelity UI at scale Dev handoff Client reporting
Legacy field portal map view, cluttered with overlapping asset labels
Before — legacy map view
Redesigned field portal map view, cleaner and easier to scan
After — redesigned map view
Works Programmer persona built with Copilot
Persona built with Copilot
Grid of dev-ready screens for the field portal
Dev-ready screen set (170+ screens)

"Irwin played a lead role in enhancing the usability and design consistency of the Field Portal during an intensive 8-day engagement... I was impressed with your approach, you delivered a great outcome for the client that no doubt will lead to further engagements."

— Associate Creative Director, Experience
Pitch & new business · Member services

Winning a pitch with a working prototype instead of a deck

Client: one of Australia's largest member-owned mutual organisations, providing roadside assistance, travel, and insurance services. Brief: support the tender response covering road, travel, and rewards services.

EngagementPitch / pursuit
RoleDesign & prototyping contributor
Output2 working AI prototypes
Tools used
Microsoft Copilot Figma Make Figma

What happened

  • Built a functioning member-archetype agent in under an hour using Copilot, drawing on documents from previous engagements to ground it in real customer context.
  • Rapidly prototyped an interactive, customer-facing orchestration agent in Figma Make in roughly two hours, demonstrating how the proposed solution would capitalise on key business opportunities.
  • Both artefacts started as personal experimentation but so closely embodied the proposed solution that they were folded directly into the final pitch.
  • Was asked to present both deliverables directly to the client's executive team — producing presentation-quality outcomes without senior direction during creation.

Skills exercised

Executive presentation Rapid AI prototyping Conversational agent design Pitch & sales support Solution storytelling Working under ambiguity
Member archetype agent conversation built with Copilot
Member archetype agent (Copilot)
Orchestration agent prototype built in Figma Make
Orchestration agent (Figma Make)
Live demo — the orchestration agent prototype in action

"In a very short timeframe, he built a working member archetype prototype agent, which quickly became a core asset in both the proposal and the pitch... his live demonstration became a defining moment in the pitch, a genuine 'wow' factor — reflected in the CEO directly commenting on how impressed she was with the demo."

— Senior Designer, CX&I, Experience

"These two absolutely nailed it! First pitch in the bag, CCO interrupted the demo and said 'let's just build it.'"

— Group Creative Director
AI product uplift · Urban planning

Sharpening an AI planning tool inside a two-week sprint

Client: an urban planning consultancy. Brief: enhance the UX of an AI proof-of-concept that automated a manual planning-permit process, cutting what had taken weeks down to hours.

Engagement2-week sprint
RoleUX/UI lead, cold start
Output6 enhancements shipped, 2 scoped
Tools used
Figma FigJam

What happened

  • Joined with no prior exposure to the client or the proof-of-concept, and needed to ramp up fast enough to withstand executive scrutiny.
  • Independently led a UX/UI workshop with 22 senior urban planners to identify and prioritise enhancements.
  • Translated workshop feedback into high-fidelity wireframes and a working prototype inside the two-week window.
  • Delivered six major enhancements for the current phase, plus two more scoped for the next.
  • Presented deliverables directly to senior executives, including a planning partner, the CIO, and the Head of Data and Applications.

Skills exercised

Large-group workshop facilitation Fast ramp-up on unfamiliar domains Executive presentation Wireframing under deadline Prototyping Autonomous delivery
Workshop prioritisation board from the uPlan UI workshop
Workshop output & prioritisation board
Before and after comparison of the planning dashboard
Before / after — planning dashboard
Final AI-powered planning assessment prototype screen
Final prototype — AI assessment generation

"Despite having no prior experience with the client... Irwin quickly understood the subject matter and delivered a functional user interface prototype within two weeks, which was well received by the planning SMEs. I was particularly impressed by his facilitation of a workshop with senior urban planners."

— Manager, Strategy and Consulting

More case studies (including public-sector education and automotive work) are in progress — get in touch if you'd like to see additional examples.

Ask for more examples