Move your mouse to paint, or press the buttonSwipe to paint, or tap the button
Interview · 27 questions · my AI workflow
These are the questions I’m asked most for senior and lead design roles, answered the way I’d answer them in the room. Each answer points at work on this site, or says it’s judgement. Where I don’t have evidence, I say so. For the full method, see how I work.
Showing 27 of 27 questions
Ten years in fintech, SaaS, Web3 and real estate. On product teams at Mara and Postpay, then Toptal clients, and now Zedo, where I lead three contract designers. I turn messy problems into flows people trust, and I care about the revenue those flows protect.
About 15 seconds
The longer version
The thread through all of it is trust, because trust is what turns a visit into revenue. At Postpay it was showing the whole repayment commitment on one screen. At Mara it was explaining fees and verification to people who had never owned crypto. At Zedo it’s showing which listings are verified. I’ve also led people: I lead three contract product designers at Zedo, a junior designer worked under me at Postpay, and I’ve given 538 one-to-one mentoring sessions on ADPList. Now I want a lead or head-of-design role where I set the direction, tie design to the numbers the business cares about, and still open the file.
My Toptal engagements finished in October 2026. I’d rather commit to one product and one team than keep moving between projects, and I want to lead designers as well as design. I’m choosing deliberately, not in a hurry.
About 15 seconds
The longer version
Consulting taught me to find the real problem fast, on ProConnection, TurnKey Coach and Outdoor Living Supply. What I miss is the long loop: ship, watch what happens to the numbers, go back and fix it. A one-off audit can’t give me that. I’m open to full-time, contract or advisory work, and the best fit is a company where design has a say before the roadmap is fixed, and where someone can tell me which business number my work is meant to move. I run Zedo alongside, so I’m not looking out of necessity.
Before I talk to a company I use its product, write down three things I’d change and one I wouldn’t, and read what its leaders have said in public. I’ll tell you which made me apply, which beats a line about mission.
About 15 seconds
The longer version
I already do this for companies I haven’t met. My consulting page has teardowns of Wekeza, Cluely and Careerspan: what I looked at, what I found, and what I don’t know. I rank findings by severity, so you can see what I’d fix first and why, not only what I noticed. If you give me an hour with your product and the one number you care about, you’ll get the same: three changes, one thing to leave alone, and how I’d tell whether it worked.
Leading a design team small enough that I know each person, and close enough to the work to review it, with a clear line from what we design to the numbers the business reports.
About 15 seconds
The longer version
That’s the direction my last few years have pointed in. Running my own product has taught me to answer for trust and cost as well as screens, and I lead three contract designers. Outside work, I’ve spent the time I could spare mentoring other designers on ADPList. The Heirs ERP showed me the next step: a platform for organisations with more than 70,000 staff needs a design team that works as one system. I want to do that inside a larger product organisation, with real scale, and be judged on what changes for the business.
I start with the decision the team has to make, not the screen. Then I find out what’s already known, draft the smallest flow that could work, test it with real users, and hand over with the reasoning, not only the files.
About 15 seconds
The longer version
J2U is the recent example. I audited the market leader first, reading its public feed of 1,103 products, then wrote a strategy document the founder approved, designed 27 page boards, and kept one build document that every working session started from. For Chotcut I started from the goals: delivery within 60 minutes, a wider market for supermarkets, a cashless economy. Then came five personas and their journeys before any screen. For the Heirs ERP I wrote the plan in the order the dependencies run: foundations, global modules, HR, then handover.
Zedo, because I carry it end to end: research, trust model, product and the business around it. It’s live in Lagos, Abuja and Port Harcourt with about 1.4K users, and it’s the one where my mistakes are visible.
About 15 seconds
The longer version
I’d rather show where my own product misleads people and how I’d fix it than tell a clean story. The case study lists three: “Close” is too generous in the Worth a look list, missing data is shown as if it were data, and an unfinished listing loses its photos. Next to those it shows what’s live: plain-language AI search on web and WhatsApp, verified agents, an admin console that runs verification and listings, and private paid Circles. I like that a reader can check my decisions against a real product, not a mock-up.
I write down what I know, what I’m guessing and what would change my mind, then pick the option that’s easiest to reverse. I label which claims are evidence and which are judgement, on this portfolio too.
About 15 seconds
The longer version
At Postpay I treated three moments as the product: whether a shopper completes checkout, whether they understand and repay, and whether they come back. That was my judgement, and the page tags it as judgement, not a finding. At Chotcut I would test the riskiest sentence in the goals, the 60-minute delivery promise, with a handful of supermarkets, a handful of riders and a paper order sheet before designing screens around it. Both are the same habit: find the claim that costs the most if it’s wrong, and test that first.
I decide the measure before I design: completion, time on task, wrong answers, support questions. And I say plainly when I don’t have results, so several case studies show what I’d instrument instead of quoting numbers I can’t prove.
About 15 seconds
The longer version
For J2U, which opens on 15 November, I’d watch four things from day one: how many carts reach the delivery estimate, how many reach checkout, how many Request from Japan forms become a paid deposit, and how many WhatsApp questions are about price or delivery. For Outdoor Living Supply the brief set the targets: sales uplift from recommended items, daily use by staff, and the share of recommendations marked useful. On Zedo I set a guardrail: click-through on the exact match must not fall. I haven’t run that test, so I don’t claim a lift.
Foundations first: colour, type, grids and spacing, then components that earn their place. On the Heirs ERP the foundations were agreed in the first two days, so a form built in week two matches a dashboard from week one.
About 15 seconds
The longer version
At J2U the brand tokens and rules were written before any template: two typefaces, a short palette, pill buttons, and a standing rule that Japanese text must be real. At TurnKey Coach I documented the system the audit produced, so developers could build from it. Lions’ Den Nigeria needed one gold-on-black identity across broadcast, social and an application page. WellnessSync runs a watch, a phone and a TV from one state and one progress ring. The point of a system is cost: rules like these stop the same argument coming back every week and cut the time spent redrawing what already exists.
As a default, not a last pass: text contrast at 4.5:1, big touch targets, keyboard paths and states named in words, never colour alone. I run axe on this site’s pages in light and dark before I publish.
About 15 seconds
The longer version
At J2U the two stock states, In Lagos and Ships from Japan, use two colours and two words, and I fixed a waitlist header that overlapped its heading on 320 and 360 pixel phones. At TurnKey Coach every row carries a worded status, Completed, Incomplete or Missed, not a dot. In the Wekeza audit the first findings were low contrast and weak emphasis on the sign-up button, so I rated them high severity and listed the ratios to meet. Accessibility is also a business argument: every shopper who can’t read the price or reach the button is a sale you’ve already paid to attract.
I use what already exists: reviews, competitor audits, support threads, a handful of short calls. At ProConnection I rated every page from customer reviews and competitor research before redesigning anything.
About 15 seconds
The longer version
At Wekeza I audited a live app against ten usability heuristics and ranked each issue by severity. I say what that is: a heuristic review, with AI as a collaborator on the first pass, not user testing. At Chotcut I wrote five personas with a journey each, scenes not statistics, and labelled them as my judgement. A cheap method is fine if you’re honest about what it can’t tell you. At ProConnection I kept the persona labelled as review-based, not interview-based, for the same reason. It still lets a team spend its budget on the riskiest question.
I start from where money is made or lost and design for those moments. At Postpay that was completing checkout, understanding the repayment, and coming back. I’m clear about which results are mine and which are a benchmark.
About 15 seconds
The longer version
At Postpay I mapped the journey into three revenue moments and built two prototypes, a checkout with exact instalment maths in English and Arabic and a lifecycle view across three surfaces, to show how I’d protect each one. At Chotcut the goals were business ones: reach more supermarkets, deliver within 60 minutes, support cashless payment. At Outdoor Living Supply the brief’s targets were revenue ones: sales uplift from recommended items and daily use by staff. I don’t have Postpay’s shipped revenue figures and I won’t quote any. I show the reasoning, how I’d measure it, and what I’d change if the number didn’t move.
I price the problem in the team’s own units first: abandoned checkouts, support questions, hours of rework. Then I compare options by cost and risk, name the measure, and report what moved. If there’s no result yet, I say so.
About 15 seconds
The longer version
I’d rather give a decision-maker a small, honest model than a big number. For Outdoor Living Supply the problem was reps retyping SKUs and looking up stock elsewhere, which slows the customer and wastes upsell chances; the measures the brief set were sales uplift, daily use and useful recommendations. For Postpay I would put analytics on checkout completion, approval rate and on-time repayment rate before shipping. I don’t have measured returns for most of this work, and the pages say so. Where I do have real numbers, I show them with a label.
By what the business loses if it waits. I put the shared foundations and the money or trust moments first, and park the rest with a reason. On the Heirs ERP the order was set by what every later module inherits.
About 15 seconds
The longer version
The Heirs plan runs in dependency order: foundations and global modules, then HR in two stages, then a full week for feedback and handover. At TurnKey Coach the first fix was the coach’s review queue, because it’s the loop the coach uses most, so Needs review now comes first with its count. At Chotcut I kept the core experience to five steps, select a store, add to cart, pay, order pickup, door delivery, and hung everything else, recipes, stores and profile, off that line. Prioritising is also saying no in writing, with the reason, so the next person doesn’t reopen it.
I don’t argue taste. I bring a question the evidence can answer and write the outcome down so we don’t reopen it. With J2U’s founder we agreed what to borrow from the market leader and what to refuse, and recorded the refusals.
About 15 seconds
The longer version
The pattern is the same everywhere: name the decision, show two options with what each costs, say what I’d measure, then record who decided and why. The cost matters most to a business owner, so I put it in their terms: launch date, support load, risk to trust. At TurnKey Coach, for example, I wrote each change as before, what we found and what changed, so the argument was about the evidence on the page. I keep names and details of specific disagreements off a public page. Ask me in the room and I’ll tell you the story.
Early, and in their language. I share the reasoning with the files: flows, states, edge cases and a component list. The Heirs plan ends with a handover pack and a session with the developers.
About 15 seconds
The longer version
At Mara I worked with whoever was on the outside product team that month, on every page, and kept the product direction consistent while the people changed. That only works if decisions live in a document, not in someone’s head. On ProConnection the redesign was handed to developers with a design system of components, styles, fonts and colours, so nobody was guessing at a hex value. Engineers also tell me what is expensive to build, and I’d rather hear it in week one than week six, because that is where schedules and budgets are won or lost.
I ask for it on the decision, not the pixels: what would make you not ship this? Then I write down what I’m changing and what I’m not, with the reason. I’m used to being wrong in public.
About 15 seconds
The longer version
My own Zedo case study lists the ways my product misleads people. It’s uncomfortable to publish, and it taught me that people trust a designer more when the flaws come first. The TurnKey Coach audit works the same way: every finding is written as before, what we found and what changed, so a coach or founder can disagree with a specific line, not a general feeling. Critique is cheapest before development starts, and I’d rather lose an afternoon to a hard review than a sprint to a rebuild.
I’m remote-first, based in Lagos, mostly working with teams in the US, the UK and the Gulf. I write things down: a one-page brief before a call and a decision log after. Async is the default, and overlap hours are explicit.
About 15 seconds
The longer version
Postpay was run from Dubai with me remote; Toptal clients were across time zones from the start. Good remote work is mostly good writing: the next person should be able to pick up the work without asking you. On the Heirs ERP the plan ended in a handover pack of files, screen map, user flows and component notes, so the developers could carry on without me in the room, across any time zone. That is also what protects delivery dates and budgets when a team is spread out.
I review the thinking, not only the output. At Zedo I lead three contract product designers, at Postpay a junior designer worked under me, and I’ve given 538 ADPList mentoring sessions, rated 4.9 from 63 reviews.
About 15 seconds
The longer version
I try to leave people with a way of working, not a list of fixes: how to frame the decision, how to show two options, how to say what’s still unknown. I also mentor through DigiFemmes, supporting women entrepreneurs in Francophone Africa. I won’t claim a large team. The honest summary is that I’ve led small, mentored widely, and I’m ready to lead more. What I’d bring is a review habit, a written way of deciding and a clear link from each designer’s work to what the business is trying to achieve.
Month one: listen and use the product, and talk to every designer, product manager and engineer. Month two: fix the one thing that blocks the team most. Month three: set a small quality bar and a review rhythm, then show a result.
About 15 seconds
The longer version
I’d resist a big reorganisation early. The fastest way to earn trust is to remove one real irritation, visibly, and be clear about what I’m not changing yet. I’d also ask the leadership team for the two or three business numbers design is expected to move, and write them at the top of the team’s plan, so every project can say which one it serves. In my audits the fix that matters is usually the loop people use most, such as the coach’s review queue at TurnKey Coach. This is judgement, not a result I can show you.
I bring a one-page argument: the decision, two options, what I’d measure and what it costs, in the business’s language: completion, revenue moments, risk. At Postpay I framed the work around the three moments where revenue is made or lost.
About 15 seconds
The longer version
Executives don’t need more screens; they need a clear choice and an honest view of what we don’t know. At Outdoor Living Supply the case I’d make for the AI assistant rests on three measures from the brief, sales uplift from recommended items, daily staff use and recommendations marked useful, and I’d put a cost beside each option. If I can’t say what would change my recommendation, I haven’t thought it through. Strategy conversations go better when design brings a number to the table, or says clearly that it doesn’t have one yet.
Systems and rituals before headcount: a shared foundation, a review rhythm and a written way of deciding. Then hire for the gaps, weighting judgement and writing as heavily as polish.
About 15 seconds
The longer version
On the Heirs ERP, where the product spans finance, HR, payroll, procurement, supply chain and operations for many organisations and roles, the plan puts shared foundations first because every later module inherits them. A team works the same way. I’d also give the business a simple view of what design costs and what it returns: which projects use the shared system, how long reviews take, what shipped. Headcount requests are easier to approve with that evidence. I haven’t built a large team, so this is how I’d start, not what I’ve done.
I decide what must be exact and what can be rough. On the Mara swap review the price impact, the minimum you’ll receive and the network fee had to be right; the chrome around them could change.
About 15 seconds
The longer version
At Postpay the instalment arithmetic had to add up to the order total to the last fils, so that’s what I built and tested first. Speed comes from being clear about which decisions are expensive to reverse. Money, identity and data deletion get time. A button’s corner radius doesn’t. I ship the exact parts first and label the rest as draft. On the Crowdcast host console, one live overlay on stage at a time keeps a host fast under pressure. That is also how I protect revenue: the places where a shopper decides to pay or leave get the care, and everything else can follow.
As a fast collaborator I supervise, not an author. It helps me read research faster, draft flows, build prototypes and write specs. I decide, I verify and I own the result. The section below walks through the workflow.
About 15 seconds
The longer version
This portfolio is part of the evidence: its prototypes were built in code, its claims carry labels, and the J2U case study says plainly that the build ran through an AI assistant on a copy of the theme that I reviewed and published myself. The assistant couldn’t publish, refund or move money. The business value is time: less spent on the legwork of audits and catalogue clean-up, more on the decisions a founder pays a designer for. I don’t quote a time saved, because I haven’t measured it.
It sounds sure when it’s wrong. It invents evidence, smooths over edge cases and flatters a bad idea. So I never let it supply a fact I can’t trace, I ask it to argue against me, and I check numbers by hand.
About 15 seconds
The longer version
Every claim on this site carries a label: public, measured, design-level, judgement, or stated by me. The labels exist because AI makes it easy to write a confident sentence with nothing under it. At Zedo the AI search carries a standing line saying it suggests from available listings, and I treat money copy as a design decision, because overpromising is the fastest way to lose trust, and lost trust is lost revenue. In a review I’d rather be caught saying I don’t know than defend a figure a model made up.
I’d like to hear the range for the level first, then I’ll tell you whether it works. I’m open to the employer’s offer and I’ll ask for the going rate for lead roles in that market.
About 15 seconds
The longer version
I’m not going to name a number before I know the scope: how many designers, how much strategy, how much hands-on work. A lead role that’s mostly hands-on and a head-of-design role are different jobs, and they’re measured differently. I’d also like to know which business outcomes the role is accountable for, because that shapes both the level and the pay. I’m happy to talk about base, equity and contract rates once the scope is clear, and I’d rather agree it in the open than negotiate around it.
I’m based in Lagos and I work remotely. I’m open to relocating when a company sponsors the move, and I won’t assume I can work somewhere I don’t have the right to. I’m available now.
About 15 seconds
The longer version
Tell me the location and the arrangement and I’ll tell you plainly what it would take. My Toptal work finished in October, so I can start without a notice period. I’ve worked with teams in the US, the UK and the Gulf, so time-zone overlap is something I plan for, not a surprise. I’d rather lose a role on honesty than win it on a promise I can’t keep, and I’d rather a hiring manager knew the constraints in the first call than found them at offer stage, when it costs everyone time.
AI workflow · 8 steps
It’s one of the questions I get asked most now, so here’s the whole workflow, step by step. AI does the legwork. I make the decisions, check the facts and own the result. Each step shows the problem, where I’ve done it, what it produced and what I watch for.
Step 01 of 8
The problem: I keep rebuilding context: why did we decide that, and what is still open?
What I do
Tools
What I watch for
Notes go stale, so every decision carries a date.
What it produced
27 page boards approved before the build, and no guessed numbers in the store.
I start from the problem and only then choose the tool. The same tool can serve more than one step, because the question is never “what does this tool do?” but “what am I trying to solve right now?”
Your turn
A company hiring a designer is buying an outcome, not a pair of hands, so my questions start with the business: the number design should move, how it reaches revenue or cost, and what a miss costs.
Which business number should design move this year, and who owns it?
So I know what my work is measured against, not only what it should look like.
How does the company connect design work to revenue, cost or retention today?
It tells me whether design is treated as a cost or as a lever.
Where does design get involved: before the roadmap is fixed, or after?
Early is where the cheapest decisions get made.
When design and engineering disagree, who decides, and how?
A clear owner saves weeks. A missing one costs them.
What was the last design decision that turned out wrong, and what changed?
How a team talks about a miss is how it will talk about mine.
What does a failed launch cost you, and how early would you want design to catch it?
It sets how much time the risky parts deserve.
What would make this role a success in six months, in numbers if you have them?
I’d rather agree the bar on day one than discover it in a review.
What do you wish a candidate had asked you?
Usually the most honest answer of the interview.
I took the question themes from public interview guides and candidate reports for product designers and design leads, then chose the ones that come up most. The Toptal wording comes from anonymous candidate reports, so treat it as a guide, not a list. The answers are mine.
(04) — Contact
Open to opportunitiesLagos, Nigeria
Design lead and Senior Product Designer · full-time, contract or advisory. Pick what fits and it opens an email with the right subject.