Fraud Detection & Response
At UOB, I designed the operational screens Fraud Ops use to monitor fraud-scored events and manage which events are sent to an external vendor for fraud scoring.
UX Designer · Singapore
Yuting
Wong.
Role
UX Designer
Focus
Corporate banking
Currently
UOB Infinity
Professional work
UOB Infinity Rebranding Initiative
I led the design direction that brought UOB’s new brand into corporate banking, adapting it for dense, high-stakes workflows.
Earlier work
Nested
My General Assembly final project, a consumer mobile concept. A home setup companion built around a renovation checklist.
UOB Infinity · Rebranding initiative
Translating a consumer brand into corporate banking software.
I led the UX direction, facilitated workshops across teams, and shaped how UOB's new brand could fit Infinity without getting in the way of how people already worked.
Refreshing Infinity without disrupting the way corporate teams work.
UOB had introduced a refreshed visual identity for its retail products, and Infinity needed to align with it. But corporate banking is very different from consumer banking. Users spend hours working with dense tables, approvals, and payment workflows, so simply applying the new design system would have made everyday tasks harder.
A rebrand introduced in retail had to work for corporate.
UOB rolled out its refreshed brand across retail banking first, and Infinity needed to adapt it for corporate banking. Although both products shared the same brand, they served very different users and workflows.
Retail banking
Corporate banking
Simple, task-focused journeys
Complex payment and approval workflows
Spacious layouts
Dense, information-rich tables
Brand experience
Speed, accuracy, and operational clarity
Applying the retail patterns directly created usability issues. Lighter hierarchy made tables harder to scan, spacious layouts separated related actions, and softer status colours reduced the distinction between important states.
The goal wasn't to recreate the retail experience. It was to bring the new brand into Infinity while keeping the product familiar, efficient, and easy to scan.
Design direction
Led the visual direction for adapting UOB's refreshed brand to Infinity's corporate workflows.
Workshops and alignment
Facilitated workshops with Product, Sales, Branding, and Trade to align on a shared design direction before detailed design began.
Stakeholder presentation
Built and presented the proposal that secured cross-functional approval for the design direction.
Design governance
Reviewed and guided the application of the new brand across dashboards, transactions, navigation, notifications, PDFs, and app icons.
Rather than redesigning everything, we focused on introducing the new brand where it improved the experience and keeping familiar patterns where they supported speed and accuracy. We explored four dashboard directions using real transaction data and reviewed them with stakeholders side by side.
ADark · High contrast
Too heavy. Increased visual fatigue in long scanning sessions.
BLight · Spacious
Important actions felt disconnected. Slower task flow.
CLight · Balanced
Best balance of brand expression and operational clarity.
DDark · Soft surfaces
Statuses lacked distinction. Higher risk of misread.
The chosen direction brought Infinity closer to the new TMRW brand while preserving the dense, operational interface analysts relied on every day.
What changed
- Introduced lighter content areas to improve readability.
- Increased contrast for transaction states in dense tables.
- Applied the new brand selectively where it supported usability.
- Preserved existing workflows so users didn't have to relearn the product.
The stakeholder presentation deck: eight frames covering the existing experience, design challenges, four visual directions, and the recommended approach across key surfaces.
Two decisions that shaped the final direction.
1. Refreshing the visual hierarchy with the new brand
The previous dashboards relied on dark blue cards that made dense information feel heavier over time. Feedback from operations users is what settled this one. They spend long sessions scanning these tables, and the dark surfaces wore them out faster.
So we moved to UOB's lighter blue palette. It reads cleaner, makes dense tables easier to scan, and still sits within the refreshed brand.
Left, the original Infinity dashboard. Right, the rebranded direction: lighter surfaces, white cards, and a dark navigation header.
2. Redesigning the progress stepper
Users couldn't quickly tell which step they were on because completed and upcoming states looked too similar.
I worked with another designer to redesign the stepper so every state had a clear visual cue. Completed steps turn green with a tick, the current step is highlighted in orange, and the progress bar fills as users move forward. I also added explicit step labels to make the sequence easier to follow.
The stepper before and after. The redesign gives the finished step a green tick, fills the bar as you advance, and labels each stage so the sequence reads without decoding circle colours.
Every stakeholder group looked at the redesign through a different lens.
TeamFocus
BrandingBrand consistency
ProductFamiliar workflows
TradeIncremental rollout
OperationsReadability and scan speed
Rather than presenting a complete redesign, I structured the proposal around an incremental rollout: what would change immediately, what would remain familiar, and what could evolve later. That approach aligned the teams and secured approval in a single review cycle.
Turning a concept into a buildable plan.
After aligning on the design direction, I translated it into a detailed implementation specification covering every affected screen, component, and interaction state. It became the shared source of truth for engineering estimates, business requirements, development handoff, and QA, helping teams deliver the rebrand consistently across the product.
The full specification, zoomed out. Every affected screen, component, and state mapped old to new for the team to build from.
The proposed direction was approved by Product, Sales, and Trade, with updates spanning dashboards, navigation, notifications, transaction flows, PDFs, and app icons.
This concept was developed as part of an internal rebranding initiative. The project reached stakeholder alignment but was not released due to changing business priorities.
Operations feedback ended up shaping our biggest decision, but it came in late. Next time I'd bring ops users into the very first workshop instead of the later reviews.
I'd also validate concepts through quick usability sessions. Most decisions came from stakeholder reviews, but observing analysts working with real transaction data would have provided stronger evidence for design choices around readability and status visibility.
UOB Infinity · Fraud operations
Fraud Detection & Response
Changes to UOB's fraud controls were being proposed and approved over email and spreadsheets, even though one change could affect many customers. I designed the screens in FDR that Fraud Ops now use to investigate fraud-scored events, change what gets scored, and get those changes approved in one place.
Under NDA
Internal UOB work. The screens are recreations with made-up data, and some names and settings have been changed.
Live in 2 markets
In daily use by fraud operations teams.
Centralised approvals
Approvals used to happen over email and spreadsheets. Now they happen in one place.
Full audit trail
Every change records who made it, why, who approved it, and when.
Overview
Every customer action in UOB Infinity, like logging in or adding a payee, is an event. Events can be sent to an external vendor for fraud scoring, but Infinity never talks to the vendor directly. Everything goes through FDR.
The problem
Fraud Ops could investigate fraud, but controlling which events get sent for fraud scoring was still manual. Changes and approvals were handled through emails and spreadsheets, creating risk when one change could affect many customers.
How might Fraud Ops go from
- 01understanding what's happening
- 02deciding what to change
- 03requesting it
- 04getting it approved
- 05keeping a record
all in one place and without side emails or spreadsheets?
How it works
How the fraud workflow works
FDR checks whether scoring is switched on for that market, channel and event. If it is, it hides customer details before sending the event out, then matches the score that comes back to the real customer. My screens sit inside FDR.
Example: a customer adds a payee
Customer
Adds a payee
On UOB Infinity
FDR
Checks the settings
Is the market, channel and event on?
No
Result
Not sent
Skips fraud scoring
Yes
FDR
Masks the details
Hides names and account numbers
External
Fraud vendor
Runs its fraud rules and returns a risk score
FDR
Matches the score
Links it back to the real customer
Customer
UOB Infinity acts
Allow, OTP, or block
FDRFraud vendorUOB Infinity
Design · 1
Monitor
Where Fraud Ops search and review scored events. The vendor only ever sees masked IDs, so Monitor is where those results get linked back to a real customer or account.
Filters can be combined, because investigations don't always start from the same piece of information. Monitor doesn't change scores or rules. It's for understanding how widespread an issue is before anyone touches a control.
Design · 2
Event Configuration
Where Fraud Ops switch scoring on or off at market, channel or event level. The higher the level, the more events a change affects, so nothing applies until a second person approves it.
Market → Channel → Event
Bar length = events affectedWidest reach
Level 01
Market
e.g. Singapore
Every channel and event in the market
Level 02
Channel
e.g. UOB SME
Every event in that channel
Level 03
Event
e.g. Add payee
Only that customer action
Higher levels affect every event underneath them.
Approval workflow
A maker changes a setting and gives a reason. A checker reviews it and approves or rejects. Every change records what changed, who asked, why, who approved it and when.
User flow
How a configuration change gets approved
A change is proposed, held pending, reviewed in place, and either goes live with an audit entry or returns to the maker.
Maker in Event Configuration
Changes a setting, writes the reason
Edited in place, same table
submit
Same person approving?
no
yes
Pending
Sits in the same table
Nothing is live yet
Checker opens the detail view
Sees what changed, and why
Checker decides
Blocked, needs a second person
Whoever proposes a change cannot approve it
approve
reject
Change goes live
Returned with a remark
Audit entry covers submission → approval
back to the maker, nothing changed
User flow
How a configuration change gets approved
A change is proposed, held pending, reviewed in place, and either goes live with an audit entry or returns to the maker.
-
Maker in Event Configuration
-
Changes a setting, writes the reason Edited in place, same table. Then submits.
-
Same person approving?yesBlocked, needs a second person Whoever proposes a change cannot approve itnoCarries on to review
-
Pending Sits in the same table Nothing is live yet
-
Checker opens the detail view Sees what changed, and why
-
Checker decidesapproveChange goes live Audit entry covers submission → approvalrejectReturned with a remark Back to the maker, nothing changed
Key design decision
Separating Monitor and Configuration
Monitor is for investigating. Event Configuration is for acting. Keeping them apart means people understand an issue before they change a control, and nobody flips a setting by accident in the middle of an investigation.
Key design challenge
What should the checker see?
My first version showed the checker the entire configuration with changes highlighted, to give full context. But most requests only touched a few items, and those got lost in the full list. So I switched to showing only what changed, which answers the checker's real question: what am I approving?
What I learned: more information didn't lead to a better decision. The right information had to be easy to find.
Trade-off
This works well for small requests. A big batch of changes still gives the checker a long list. I chose not to restrict what Fraud Ops could change, and leaned on making the impact visible plus the second approval instead.
Before and after
Before
01Investigate the issue
02Determine what needs to change
03Document the change
04Request approval through email
05Track the approval separately
06Apply the change
07Maintain the approval record separately
After
01Investigate in Monitor
02Make the proposed change in Event Configuration
03Provide a reason
04Checker reviews the change
05Approve or reject
06Approved configuration is applied
07FDR keeps the audit record
Measuring impact
What changed, and what I'd track next time
We didn't set formal metrics. It was a six-week internal tool, so the clearest signs are in how the work happens now.
Today
What's different now
- Configuration changes go through FDR instead of emails and spreadsheets.
- Every change has a reason, a second approver and a full record.
- No one can change the configuration on their own.
Next time
If I did it again, I would set a baseline at the start and track:
- 01How long a change takes from request to approval, before and after.
- 02How many changes still happen outside the tool.
- 03How often checkers send requests back, as a sign of whether the review screen gives enough context.
- 04Audit findings related to configuration changes.
What I would do differently
Six weeks meant I moved into interface design quickly. With more time, I'd test the checker view earlier with large requests, to learn how much a checker needs to see before approving with confidence.
General Assembly · Final project
UX case study
Nested is a home setup companion for anyone furnishing a place in Singapore. This write-up covers the research, the competitor teardown, the people I spoke to, and how a renovation checklist ended up running the whole thing instead of sitting off to the side as one more feature.
Nested: helping people furnish a home, not just fill a cart.
A home setup companion for people furnishing a place, built out from real interviews rather than a moodboard. It started with new homeowners, but the deeper I got, the clearer it became that the same wall trips up anyone setting up a space. A long list, no order, and no real sense of what's worth buying.
What Nested is.
Nested started as my final project at General Assembly. I chose first-time homeowners in Singapore, because a few people around me were going through it and the stress was real. You buy a flat, renovate it, and furnish it, and all three tend to land at once, often on top of planning a wedding.
Nested is a home setup companion. You browse by room, read reviews, preview a piece in your space with AR, and check out. Running underneath all of it is a renovation checklist that tracks what a finished room needs and what you still have to get.
Problem space
People setting up a home are often clueless about where to start. New owners, sure, but also renters furnishing a first place, or someone redoing a single room. They don't know what to buy, in what order, or which of it is actually worth the money.
The welcome screen, your setup progress, and the checklist that quietly runs the whole app.
First, the mess.
Before I talked to anyone, I tried to capture what a move actually feels like. Not an orderly list, but a desk that keeps piling up while you're still unpacking the last box.
$$$IKEA receipt
recheckMeasurements
ideasPinterest inspiration
which?Paint colours
will it fit?Sofa dimensions
when?Delivery SMS
keepWarranty
$$$Contractor quote
unread 42WhatsApp messages
overBudget spreadsheet
laterAppliance manuals
todoShopping list
printFloor plan
!!Calendar reminders
maybeLighting ideas
ask elec.Electrical notes
Sixteen open loops running at once. This is the state people are in before they even think about buying a sofa.
Then, the people living it
I interviewed first-time homeowners to find where all of that breaks down. Three voices shaped the project.
Weiming
Doesn't know the order
Struggles to get good information on deals and suppliers. In his words: "I don't know what items to buy that contractors need to install first."
Jiayuan
No time to travel
Spends too much time travelling between showrooms, and has to look after kids, so she'd rather do the whole thing online.
Jingxin
Can't trust reviews
Unable to find proper reviews or visualise products, and overwhelmed by "so many e-commerce mediums, I don't know which platform to buy from."
Competitor analysis
I compared the home-shopping apps these users already reach for. They all sell home products, but none of them pulls the whole job into one place. Only IKEA let you preview a piece with AR, which felt like the one thing actually worth borrowing for big purchases you can't picture in your flat.
| Brand | Clear UI | AR preview | Verified reviews | One-stop home range | Flexible payment | Returns |
|---|---|---|---|---|---|---|
| IKEA | Unclear | 365 days | ||||
| Courts | Clear | 30 days | ||||
| Harvey Norman | Clear | 100 days | ||||
| Ezbuy | Messy | 30 days | ||||
| Lazada | Cluttered | Merchant | ||||
| Nested. | Guided | Free |
IKEA, Courts, Harvey Norman, Ezbuy and Lazada, compared on UI, AR, reviews, range of home appliances, payment flexibility and returns.
Usability testing
To sharpen the pain points, I ran usability testing with two people on the competitor sites they visit most, giving each two task scenarios: cart a gas hob within a $300 budget, and add three living-room items from the category section. Two pain points came through clearly.
Pain 01
Irrelevant products
Results are cluttered with products that have nothing to do with what users are actually looking for.
Pain 02
Insufficient reviews
For the gas hobs she was considering, there simply weren't enough reviews to tell whether the product was a good investment.
From pile to pattern.
Raw
Receipts, measurements, quotes, delivery texts, about forty open tabs. All of it went up.
Grouped
Five themes. Time, trust in reviews, where to even buy, cluttered competitor sites, and one repeated wish: see it in the room first.
Underneath
One question kept coming back. What do I buy, and in what order?
Time management
Covid: unable to travel around to get homewares
Spends a lot of time navigating, can't find related items
No time to travel and has to look after kids, prefers to shop online
Insufficient reviews
Limited reviews on products, doesn't know if it's any good
Wants to know more about a product before purchasing
Too many spam messages in renovation group chats; spends ages sorting for real reviews
Limited mediums
Doesn't know where to get home appliances
Few mediums to actually purchase home products
Competitor UI
Homepage of e-commerce sites is too cluttered
Platforms like Lazada are overloaded with irrelevant listings
Suggestion
Would be a bonus to visualise items in 3D / AR before buying
Research notes clustered into Time management, Insufficient reviews, Limited mediums, Competitor UI, and Suggestions.
The one that mattered
Four of the five clusters were about friction. The fifth was the real brief: people didn't need a bigger shop, they needed to know what to get and when. That question became the spine of the app.
Who it's for.
Two personas carried the design through to the prototype. Both came out of new homeowners, but the problem underneath, not knowing what to buy or in what order, isn't unique to people who just got their keys.
S
Sharlene
32 · Financial Advisor · Singapore
"I'm all about having a fuss-free shopping experience."
Context
Very busy with little free time; shops online when she can. Recently bought a house and finds few sites that sell household products.
Goals
A hassle-free shopping experience, and to visualise how products would look in her home.
Frustrations
No one-stop platform for a wide range of household products; very few sites with AR.
D
David
30 · Career Coach · Singapore
"I'd like to see a dynamic platform that sells homewares."
Context
Newly wed, just got the keys, clueless on renovation. Not sure where to look, so he sticks to Lazada.
Goals
A platform that organises household products well, guidance on what to buy as a new owner, and more reviews to buy by.
Frustrations
Few platforms with real range; scrolling through spam in renovation group chats just to find a review.
Beyond new owners
The same wall shows up for renters furnishing a first place, or anyone redoing one room at a time. Nested is built for the setting-up problem, not for a single life event.
01
Got the keys
Empty rooms, a long mental list, and a deadline.
PainExcited, then stuck
02
Research
Google, Pinterest, friends, IKEA.
PainToo many tabs, no single source
03
Planning
Measurements, budget, timeline, room ideas.
PainInformation scattered everywhere
04
Buying
Furniture, appliances, decor, lighting.
PainReceipts and orders all over the place
05
Waiting
Delivery, installation, contractor slots.
PainWhat's actually arriving today?
06
Moving in
Utilities, warranties, manuals.
PainCan't find anything a month later
07
Living
Maintenance, wishlist, improvements.
PainNo home for the ongoing stuff
The move end to end, with the friction called out at each stage. Buying is the easy part. Knowing what and when is not.
Storyboard and user flow
From there I storyboarded one person's path through Nested, then turned it into a user flow with the decision points and the AR touchpoint built in.
Storyboard: Sharlene researches at home, finds Nested, previews a product in her kitchen with AR, and checks out satisfied.
The user flow: home → browse → product → reviews → AR → cart → checkout, with the guided checklist closing the loop.
Structure, features, mid-fi.
From the flow I defined the app's key features, mapped the information architecture, and built a mid-fidelity prototype to test the whole journey.
The call that shaped the structure
My first instinct was the obvious one, and it was wrong.
First instinct
Build a better store
Browse by category, cart, checkout. The exact thing Lazada, IKEA and Courts already do, only smaller.
What the research said
Build a plan, not a store
People didn't need more aisles. They needed to know what each room takes, and in what order. So a room-by-room checklist, phased to the renovation, became the centre. Shopping happens inside it.
v1
Category store. Browse-first. Loses to the big players on range and price.
v2
Setup companion. Checklist as the spine, phased by what happens when: demolition, kitchen, bedroom.
Where it landed
Home shows progress, the checklist drives it, and buying an item ticks it off.
Key features
01
Guided to-do list
Tells you what to buy, in the order your renovation actually needs it.
02
Wide range of categories
Home products from kitchen to living room, organised in one place.
03
Reviews section
Reviews pulled up front, so people can tell if something's worth the money.
04
Augmented reality
Preview a product in your own space before you buy it.
Sitemap
Nested.
Onboarding
Log in / sign up
Set up your home
Home · progress
Setup progress
Up next
Recommendations
Checklist SPINE
Phases (demo, kitchen…)
Task detail
Linked products
Mark done
Shop
Categories
Product detail
Reviews
AR preview
Cart & checkout
Bag
Checkout flow
Order confirmed
Profile
Account
Purchase history
Saved AR projects
Help
Information architecture, reorganised around the setup journey: onboarding seeds the checklist, the checklist drives Home, and shopping, cart and profile support it.
Mid-fidelity prototype
The mid-fi flow, screen by screen: home and category browsing through product, AR, cart and checkout.
The renovation checklist is the spine.
Every other screen hangs off it. Most home apps drop you into a catalogue and leave the planning to you. Nested works the other way. You start from a room, see what a finished version of it needs, and the app keeps track of the rest as you go.
01
Grouped by room
Pick a room and the checklist lays out what it takes to finish, item by item, so nothing gets missed halfway through.
02
Tick what you already have
Check off what's there and the list shrinks to what actually matters, so you're not buying the same thing twice.
03
What's left becomes the plan
The remaining rows are your shopping list, in a sensible order instead of a random pile of open tabs.
04
A whole home or one room
Start from a template or build your own list. That's what keeps it useful past move-in week, whether you're furnishing a flat or just redoing the kitchen.
Shopping and the checklist stay in sync. Add a hob and it ticks off the kitchen list; finish the order and the room moves closer to done.
The whole thing, screen by screen.
Setup and browsing sit inside finishing your home. You come in through the checklist, and everything else, browsing a room, previewing a piece in your space, checking out, branches off it.
Set up and browse
See it in your space
AR sits where it belongs, one tap from a product, for the big pieces worth checking against your actual kitchen before you commit.
Bag and checkout
Splash through to order confirmation, on Nested's pink system.
Nested was the first thing I took all the way from talking to strangers about their flats to a working prototype. Every part of it was mine to figure out.
The part I'm proudest of isn't a screen. It's that the checklist ended up as the spine of the product instead of one feature sitting politely next to the others.
What I didn't do was put those checklist templates in front of people who were actually setting up a home, and that's the thing I'd want to know most. The default lists are still my best guess.
About
Yuting
Yuting
Wong.
I'm a UX Designer at UOB Infinity in Singapore, working on corporate banking. Most of my work is fraud operations and payment workflows, where people spend hours in dense screens and a misread status can hold up a real payment.
I like taking those screens and making them quicker to read without making people relearn how they work.
Outside of work, I'm mostly hanging out with my cats or out on the bike.
Get in touch