Product design
I turn business requirements and user needs into clear journeys, workflows, prototypes and interface designs, from new product concepts to improvements for existing platforms.
Hey, I’m Aymen. Senior Product Designer in Riyadh.
I work across product strategy, interaction design and design systems for government, finance and technology. My focus is complex, data-heavy products, including Arabic RTL and English LTR experiences.
I’m a Senior Product Designer based in Riyadh with 10+ years across product, service and visual design.
I started in graphic and motion design before moving into product design in 2018. Since then, I’ve worked on banking, insurance, government, PropTech and AdTech products.
My work centres on design systems and complex, data-heavy workflows, including Arabic and English experiences.
I design digital products that make complex information and workflows easier to understand and use.
I turn business requirements and user needs into clear journeys, workflows, prototypes and interface designs, from new product concepts to improvements for existing platforms.
I build foundations, semantic tokens, reusable components and responsive patterns that help teams create consistent products across Arabic and English experiences.
My background in branding, graphic design and motion helps me shape stronger visual identities, clearer communication and more considered product experiences.
Some projects begin with an idea. Others begin with a product that already has users, constraints and history. I adapt the process to the problem, available evidence and project access rather than forcing every project through the same checklist.
I clarify the brief with the people responsible for the product: the problem, intended users, business goals, constraints, risks and what success should mean.
Typical outputsProject brief, stakeholder questions, assumptions and an initial design plan.
The research depends on what the project already knows. I may review an existing product, speak with stakeholders or subject-matter experts, examine available findings and analytics, or compare relevant products. When evidence is missing, I make the gap visible instead of presenting assumptions as facts.
Typical outputsResearch findings, product audit, benchmark and open questions.
I turn the evidence into a shared view of the product. Depending on the problem, this can include journeys, scenarios, service flows, information architecture or data maps.
Typical outputsProblem definition, journey or service map, requirements and prioritized opportunities.
Sketches, flows, wireframes and prototypes help the team compare behaviour and trade-offs before committing to detailed screens. Testing is used when the project provides access to the right participants; otherwise, prototypes support stakeholder and technical review while validation needs remain explicit.
Typical outputsConcepts, user flows, wireframes, prototypes and decision notes.
I develop the selected direction into complete flows, screens, states and reusable patterns. Sometimes I work within an existing design system; sometimes I establish the foundations and components the product needs. For bilingual products, Arabic RTL and English LTR behaviour are considered from the structure onward.
Typical outputsInterface designs, responsive behaviour, prototypes, components and documentation.
I involve engineering before handoff to check feasibility and clarify behaviour. During implementation, I work with developers and testers to resolve gaps and review the result against the intended experience.
Typical outputsSpecifications, implementation clarifications, design review and an updated backlog.
Detailed stories showing the problem, my role, the decisions I made and how the work developed.
Product case study
A native-advertising platform connecting campaign creation, approvals, reporting and publisher implementation.
Product and visual case study
A responsive organic-shopping concept built around seasonal and personal produce boxes.
Product case study
Partner onboarding and lifecycle management across two AdTech products, from registration to review, edits and payouts.
Product case study
A bilingual financing journey from eligibility and company authorization to application tracking and disbursement.
Product case study
Case review, specialist reporting and cross-division handoffs for industrial financing, organized around roles.
Product case study
A shared foundation for client and internal products: accessible tokens, documented components, bilingual templates and a governance model.
Selected work beyond the featured case studies, spanning multilingual public education, enterprise finance and customer-experience transformation.
UX/UI and visual-identity design for four Central Bank of Mauritania websites in Arabic, French and English. I worked with content and delivery teams on information architecture, RTL/LTR navigation and engineering handoff.
A nine-project customer-experience concept prepared for executive review. I mapped how existing teams coordinate client care, defined where AI assistance should stop and hand off to people, and created a measurement framework to connect service quality with operational ownership.
UX/UI work for internal banking and insurance applications. I translated institutional requirements and business rules into user flows, interface proposals, prototypes and prioritized recommendations for future product scope.
Earlier independent work across responsive web design, visual identity and editorial experiences. This craft foundation continues to shape the clarity and finish of my product work.
Product design since 2018, after earlier roles in graphic design. Open a role to see what I did.
Communication with Aymen was always clear, proactive, and solution-oriented. What stood out most was his strong understanding of both user needs and product goals, consistently translating requirements into thoughtful, impactful designs while challenging assumptions constructively. A reliable and talented partner to work with.
Aymen is an exceptionally talented UX/UI designer who consistently put ideas into intuitive, visually experiences that truly put users first.
In my experience working with Aymen, I was impressed by his creative vision and ability to translate it into impactful designs. He consistently delivers high-quality work on time and thrives in a collaborative environment. If you're seeking a talented designer who can bring fresh ideas to the table, I highly recommend connecting with Aymen.
Aymen was pivotal to our project's success at EY, leaving a lasting impression on stakeholders and clients alike. He's exceptionally detail-oriented, brings strong Figma skills and great visual taste, and follows a disciplined, research-driven process. A skilled and dependable designer who raises the quality of both the work and the collaboration.
Aymen is a detail-oriented designer who understands client needs and translates them into thoughtful, well-crafted designs. He's disciplined, organized, and reliable. Collaboration with him is smooth, well-coordinated, and always creates a positive, productive atmosphere.
Working with Aymen has been a great opportunity to collaborate with such a talented and professional designer. He is also known for being very generous in transmitting his skills and knowledge to the team. Keep it up Aymen!
It was such a pleasure working with Aymen. He is a talented UI/UX designer, a professional communicator, and disciplined. He pays close attention to detail, delivers quality work, and brings creative ideas to solve problems. I would love the opportunity to work with him in the future, and anyone would be lucky to have him on their team.
Aymen Zarrouki's UI/UX design expertise has been pivotal in developing ArabyAds' cohesive design system. His designs are intuitive, enhancing user engagement and product consistency. Aymen is a visionary with a user-centered approach, ensuring seamless design experiences. His leadership in the design team fosters innovation and collaboration. I confidently endorse Aymen for any advanced product design role.
Working with Aymen means growing as a designer. Beyond his strong UX expertise, he leads with kindness and patience. He has a rare ability to turn complexity into clarity while mentoring and uplifting those around him. Strategic, thoughtful, and genuinely supportive.
Aymen demonstrates an exceptional command of UX principles, combining analytical precision with strategic vision. He translates complex challenges into clear, impactful solutions with integrity and patience. What sets him apart is his commitment to collective growth: he doesn't just deliver excellence, he elevates everyone around him.
Ayman is a collaborative and supportive UX designer who always supports the team and consistently shares design trends, tools, and his own knowledge with the team. He is highly committed to his work and always delivers high-quality results. He actively seeks smarter, more efficient solutions, and his positive attitude makes him a great team player.
RevGate and Boostiny are two AdTech products at ArabyAds that work with partners such as publishers and influencers. The project began with registration and grew into the whole partner relationship: sign-up, profile completion, review, agreements, changes and payments.
I designed the partner-facing flows and the team-facing screens together, so that what a partner does and what a reviewer sees stay connected.
This started as a registration project. It did not stay one.
A partner who signs up to RevGate or Boostiny has only started a relationship. They still have to complete a profile, be reviewed, sign an agreement, change details later and be paid. The people reviewing them need to see where each partner stands and what to do next.
Looking back, that is how I would frame the problem: registration was one stage in a longer partner lifecycle, and a good sign-up form could not fix a lifecycle that the team-facing screens did not support. This is my retrospective reading of the Figma file, not a research finding.
The file holds the older screens next to the redesigned flows. It covers staged registration for individuals and companies, profile completion, review statuses, alternative payment methods, partner lists, agreements and e-signing, edit requests, flags, activity history, comments, and partner profiles with payout and commission tabs.
The Figma file proves the design scope and the experience I proposed. It does not prove release or results. I no longer have access to the original research, analytics or delivery tickets, so this page reports decisions, not outcomes.
Most of the screens only make sense once you see the states behind them. This is a simplified model I drew from the status badges, cards and filters in the Boostiny screens. It is a reading aid, not an official specification.
Main path
Side states a partner can be in
The first stage asks only for what is needed to create an account: whether the partner is an individual or a company, their name and contact details, and an optional referral. A six-step progress bar shows what is still ahead: basic information, work experience, social media, preferences, finish.
Payment details, documents and password come in a separate Complete Profile stage. It has a progress ring, a short description for each step and a time estimate next to it. The bank form lets a partner enter a Swift code and fetch the bank details, with short tooltips on the fields that usually cause mistakes, such as the account holder's full name.
My reasoning, stated as hindsight: someone deciding whether to join should not need their bank details at hand to open an account. I did not test that assumption, and I would test it first today.
A partner's status is not a label added at the end. On the partner side it shows as progress through the profile. On the review side it becomes the main way to organise the work.
The Boostiny onboarding list opens with counts for new registrations, rejected registrations, pending password creation, pending profile completion, profiles in review and approved partners. The flagged and edited profiles are counted inside the review and approved groups. Each row carries a status badge, and the row menu offers the next action: approve or reject a registration, or contact the partner by WhatsApp or email.
Colour does part of the work here, which is where my visual-design background helps. Blue marks something new or under review, orange marks something waiting on the partner, green marks approval and red marks rejection. That lets a reviewer scan a long list without reading every badge.
Partners can change their details after they are approved. When they do, someone has to see the change, check it and decide. I treated that as one flow with two sides.
On the review profile, a section shows a count of edits, and the changed fields carry a small orange marker with a note saying who requested the change. The reviewer can flag a field, open the activity log, chat with the partner or leave a comment. The agreement sits beside the profile with options to download, upload or send for e-signature.
The activity log lists each change with the old value, the new value, who made it and when. That matters most for payment details, where a reviewer needs to see exactly what moved from one method to another.
Contracts, flags, suspension, account-deletion requests and payouts are the cases that usually get added last. Here they are part of the model from the start.
The filter bar lets a reviewer isolate suspended partners, requests to delete an account, requests to update a profile, flagged profile information and agreements still pending. Each partner profile has its own tabs for global payouts and commissions. I did not design these as exceptions bolted onto a happy path. They are states a partner can be in, and each has a place to be found.
RevGate and Boostiny serve different partners, so I did not make them identical. I shared the parts that should feel the same: the profile structure, the section menu, the form fields, and the edit button.
The differences are deliberate. RevGate's partner-facing profile lists a Password section and has no review tools. Boostiny's review profile lists Work Experience and adds status, agreement, comments and the activity log. The same layout carries both, so a partner and a reviewer looking at the same record see the same sections in the same order.
The Figma file preserves a connected design for the partner lifecycle in two products:
I worked closely with product and delivery colleagues on it. I am not claiming that every area was built or released, or any measured result.
The most useful thing I took from this project is to start from the lifecycle, not the first screen. The registration form was the visible problem. The lifecycle, and the people who had to run it, was the real one.
I would do three things differently. I would record the reasons behind each decision while the project is live. I would test the split between account creation and profile completion with real partners. And I would agree the measures before design starts.
Applying for industrial financing requires more than completing a form. A company has to understand whether it qualifies, prove who is authorized to act, assemble financial and technical evidence, respond to reviewers and follow the application across several SIDF teams.
The challenge presentation reported that clients took about 25 days on average to submit a loan application, while missing documents represented 54% of rejection reasons. I designed a bilingual Client Portal concept that turns this fragmented journey into a guided service, from understanding requirements before registration to responding to review requests and managing disbursement after approval.
Two evidence sources in the project presentation set the baseline.
Source: SIDF Process Mining Analysis, 2026. Presentation-reported baselines; the denominator for the 54% is not defined.
SIDF Client Satisfaction, 2024/2025 documented four recurring client pain points.
Four client pain points, SIDF Client Satisfaction, 2024/2025
The same process analysis also showed internal consequences: repeated preliminary-check SLA breaches, rework caused by requests for additional data and low pre-screening pass-through. The client and employee experiences were not separate problems. Missing guidance and evidence upstream created rework and delay downstream.
That led to the central design question: how might we help a company understand, prepare and complete the right application evidence, while keeping both the client and SIDF teams clear about progress, responsibility and the next action?
The work combined business and client interviews, data analysis, How Might We framing and sprint questions. Solution development used journey mapping, sketches, application demos and testing.
How the challenge was framed
The findings were synthesized into four connected pillars. They became the structure of the product, not a set of marketing messages.
Four experience pillars
The journey begins before login. Potential applicants can check basic eligibility, review financing requirements and understand a typical application timeline before creating an account.
This changes the first interaction from "start a form" to "decide whether this service fits and prepare for it". It addresses late discovery by exposing expectations before a company invests time in the full application.
After submission, the Application Tracker replaces the vague state of "under review" with a visible evaluation pipeline. It shows the current stage, recent activity, responsible division and pending client action.
The design does not promise an exact decision date. It provides something more credible: what has happened, what is happening now and who needs to act next.
The concept separates personal identity from company authority. A verified login establishes who the user is. It does not automatically establish permission to submit financing information for a company.
Identity and authority as separate layers
Company linking, roles and consent make the relationship explicit. The presentation also proposes integrations with sources such as Saudi Business Center, Wathiq, GOSI, Nafath, SIMAH, SPL, MODON and Qawaem to reduce repeated entry and validate records. These are proposed, not implemented.
The feasibility-study concept would use profile data, company records, eligibility responses and proposed government integrations to prefill part of the experience. The presentation describes a 60% prefill ambition. I treat it as a prototype target, not a measured result, because no implementation or evaluation method is supplied.
The design goal was not automation for its own sake. It was to reserve the client's effort for information that only the client can provide.
The financing application is organized into five stages: project, marketing, technical and financial information, then review and submission. This gives a long assessment visible landmarks and recovery points. Clients can save progress, return later and see which sections or documents still need attention.
In the prototype, feedback takes the form of progress indicators, document status, inline guidance and contextual assistance. The interface does not wait until submission to reveal missing work.
When SIDF needs clarification, the request stays attached to the relevant section. The client sees the question, the original response and the place to submit an update in one context. Review becomes a visible loop with memory, instead of a chain of emails.
The concept includes an inbox, contextual reviewer requests, chat and appointment booking, so clients can reach SIDF without losing the application reference or the reason for the conversation. Voice, video and live availability are prototype concepts. Their value is not the channel. It is that clarification stays attached to the case.
Loan-level views group eligibility, released and remaining values, claims, supporting documents and claim status. The same company context, communication model and progress language carry from preparation through financing administration.
Arabic and English were treated as structural variants. Navigation position, reading order, alignment, directional controls and mixed Arabic and Latin content were considered for right-to-left behavior, not translated at the end.
The presentation also frames responsive access as part of the service vision: status, notifications and quick actions should stay available without a desktop session. The prototype shows the intended direction. It does not establish formal accessibility compliance, which would need dedicated testing.
The Client Portal improves the quality and visibility of information entering the service. The Internal Portal organizes how SIDF teams review that same case.
| Client side | Internal side |
|---|---|
| Submit a stage | Review the same stage |
| Missing document | Explicit document request |
| Clarification needed | Update returns to the affected field |
| Internal handoff | Client-facing status |
This is the broader product story: not two dashboards, but one financing journey seen from both sides.
The challenge presentation includes four outcome figures: 60% lower time on task, 65% lower rework, 74% task success and 71% task completion. The supplied material does not include the test method, participant count, baseline, task definitions or calculation, so I do not present them as validated outcomes.
What the evidence supports:
The work does not prove production deployment, live integrations, sustained adoption or business impact.
The strongest design decision was to treat clarity as part of the service, not as explanatory text added after the workflow was built. The experience became more coherent when it answered four questions continuously: is this service relevant to me, what do I need to prepare, what is happening to my application now, and what do I need to do next?
That same clarity helps the internal process: better-prepared applications, earlier evidence checks and contextual requests give work fewer reasons to stall between teams.
An industrial-financing application moves through several SIDF teams before a decision can be made. Each team needs different evidence, produces different analysis and may need to return work for clarification or approval.
I designed a role-based Internal Portal concept that organizes this work around the case: from initial review and evidence checks to specialist reporting, team-lead approval and the next division handoff.
The internal process is not a single queue. A case may be newly submitted, waiting for intake, missing client evidence, under specialist analysis, returned for revision, awaiting team-lead approval, blocked between divisions or overdue.
Seven representative case states
Different employees need different actions. An analyst authors a report. A team lead reviews sections and requests revision. A manager needs visibility across workload, dependencies and handoffs.
The design challenge was to make responsibility and state visible without collapsing every role into a generic administration dashboard.
The challenge presentation's SIDF Process Mining Analysis, 2026 reported three internal signals.
Source: SIDF Process Mining Analysis, 2026. As reported in the challenge presentation; units, period and denominator are not defined.
The presentation does not define the unit, period or population behind the two multipliers, so I treat them as reported process indicators. I have not converted them into percentages, reductions or causes.
Together with the client-side findings, they describe one connected service problem.
The Client Portal works on the information entering the process. The Internal Portal makes responsibility and progress visible inside it: a three-way pre-screening gate, eligibility review, assignment, evidence completeness, section-level report review and cross-division handoffs.
The work followed four phases.
| Phase | What I did | Main output |
|---|---|---|
| Discover | Reviewed the client journey, internal roles, report requirements, user stories and prototype behavior | Role and workflow inventory |
| Define | Modelled states, responsibilities, dependencies and handoffs | Role matrix and case-state model |
| Develop | Designed role-specific navigation, task views, report workspaces and approval patterns | End-to-end interactive prototype |
| Evaluate | Exercised scenarios across employee roles and audited state, permission and content consistency | Workflow corrections and prototype-quality evidence |
Process, with one artifact per phase
The project evidence supports workflow analysis and prototype evaluation. It does not establish contextual interviews or usability testing with SIDF employees, so operational assumptions remain hypotheses for later validation.
I converted the workflow into ten states, from new submission to final assessment. The model includes return, blocking and reassignment paths. Exceptions are first-class states, not notes hidden inside a file.
Case states, with return and revision loops
The interface follows the kind of work a person is doing, not their department.
Work types
The RM Division Cases view separates new submissions, applications under review, returned work, division reports and final assessments. Counts, filters and status labels help a team lead scan for exceptions before opening a case.
This turns the queue from a passive list into a view of required decisions.
A new application first passes through a structured pre-screening view. The reviewer sees the company, requested financing, ownership, financial position, eligibility indicators and client history in one summary before deciding whether the case should proceed.
Separating intake from later specialist work prevents downstream teams from starting before a case has passed the initial gate.
Internal review mirrors the sections the client submitted: project, marketing, technical and financial information, plus supporting documents. Reviewers inspect each section without translating the application into an unrelated internal structure. The case remains the common reference point across roles.
The Document Completeness Engine groups requirements by application stage and shows whether each item is submitted, in progress, missing or optional. Required evidence can block the handoff at the correct point, and the reviewer can request missing documents from the same context.
Optional items are shown with a neutral label, so they do not read as blocking failures. This is a guided workflow concept, not proof of an automated production integration.
A case can look complete inside one team while still waiting elsewhere. The dashboard combines the pending action, overall report progress, division status and current owner.
This gives employees a shared view of where the case is waiting and what must happen next.
The FTR workspace divides the report into reviewable chapters and sections. Each section carries its own status, and final approval stays unavailable until all required sections have been reviewed.
Section-level review makes revision more precise than returning an entire report without context, and it keeps authoring separate from team-lead approval.
The Sector Study Library makes shared studies visible with states such as draft, in review, pending approval and approved. The concept reframes specialist research as a reusable institutional asset rather than work buried in one case file. Approval ownership stays explicit.
Client messages carry the application reference and remain visible to the team working on the file. Calendar availability gives the client-facing employee a controlled way to offer phone or video appointments.
Together, these concepts reduce the risk of clarification requests becoming detached from the application and its evidence trail. Both are concepts: whether messages are official case evidence or a convenience channel would need a policy decision.
The employee sign-in was drawn in English and Arabic as a structural adaptation, not only a translation: reading order, brand placement and sign-in priority mirror for right-to-left use. Sign-in is a concept in the prototype, not a deployed single sign-on integration.
This page shows the English version only. I am keeping the Arabic comparison out until the mirrored layout has been redrawn and checked line by line.
| Early issue | Analysis | Design response |
|---|---|---|
| Cases appeared in one general list | Different states require different decisions | Added operational groups, counts and filters |
| Internal review could diverge from the client submission | Reviewers needed shared structure | Mirrored the client application stages |
| Missing documents surfaced late | Handoff needed an evidence gate | Added completeness states and blocking rules |
| Author and reviewer actions were mixed | The roles carry different accountability | Separated submission from approval |
| Entire reports could be returned without precision | Review needed section-level context | Added per-section status, comments and revision |
| Division ownership was difficult to see | Local completion did not equal case progress | Added cross-division handoff visibility |
Scenario walkthroughs exercised responsibilities across the prototype. A documented quality sweep covered approximately 490 prototype screen states across nine personas, with 74 automated checks passing at that point in development.
Prototype coverage evidence, not usability or production performance.
The review looked for wrong-role actions, missing empty or blocked states, inconsistent statuses, broken navigation after role changes, evidence gates that could be bypassed and fictional information that looked too realistic for public use.
I corrected the first items in the screens shown here. The others are the next design work: one case-context header, chapter-level collapse in the FTR, and a shared state model with a defined owner and next action for each status.
The portal was not released. Workload data, notifications, integrations, AI generation and live handoff status are proposed or simulated unless separate implementation evidence becomes available. Task-success and time-on-task figures that appeared in a presentation are not used here, because the sample, method and baseline are not documented.
The central lesson was that an internal tool should begin with responsibility, not features.
The useful distinctions were between kinds of work: reviewing, assigning, completing, returning, approving and handing off. Once those responsibilities were explicit, the interface could show the right action to the right role while keeping the case as shared context.
Before production, I would validate the concept through contextual interviews, observation of a real cross-division case, task-based usability tests, permission and audit-trail review, and measurement of queue age, return loops and waiting time between divisions.
SIDF serves several audiences across public services, client applications and internal review workflows. Those products need to feel related, but consistency cannot depend on designers recreating the same decisions in every file.
I helped define a design-system direction that connects brand foundations, accessibility, tokens, reusable components, bilingual templates and a governance model. It is more than a visual library: it sets out how a team can move from a recurring product need to an approved, documented pattern.
As digital products grow, small differences become structural problems. The same action gains several button treatments, statuses use different colors, layouts drift, interaction states are missed and Arabic adaptations are handled at the end.
For SIDF the problem was larger than visual consistency. The system needed to support:
The objective was therefore not to make a component library. It was to create a reusable product language and a process for maintaining it.
| Phase | What I did | Main output |
|---|---|---|
| Discover | Reviewed existing product patterns, repeated UI, inconsistencies, platform needs and external design systems | Audit themes and benchmark principles |
| Define | Set system goals, design principles, architecture and the relationship between foundations, tokens, components and templates | Design-system model and roadmap |
| Develop | Shaped accessible foundations, semantic token logic, interaction states, component organization and bilingual templates | Reusable system layers and application patterns |
| Validate | Checked contrast, responsive behavior, language direction, states and use in realistic product screens | Corrections and documented constraints |
| Scale | Defined request, review, versioning and documentation steps | Contribution and governance model |
Five phases of the design-system work
The available evidence supports a product audit, a benchmark review, accessibility work and artifact-level validation. It does not include raw interview notes or usability-test findings, so I do not present those activities as completed research.
The first step was to understand where inconsistency was coming from. I reviewed existing product UI as a set of repeated decisions rather than isolated screens:
This audit reframed the project. A design system would only help if it solved recurring product problems and reduced future duplication. It could not be a decorative inventory detached from the products using it.
I compared the planned system with established references: Material 3, Atlassian, IBM Carbon, Microsoft Fluent, eBay Evo and Uber Base/Core. The goal was not to copy their visual language. I studied how mature systems handle:
The benchmark produced design criteria for SIDF: the system had to be rational, human-focused, consistent, cross-platform and able to evolve in the open.
I organized the work into five connected layers, so the system would not be a flat library.
Five connected layers
This structure keeps a visual primitive, an interactive component and a workflow pattern separate, instead of treating every screen as a pile of small parts.
These principles became evaluation questions. Does this new component solve a need that an existing pattern cannot? Is its focus state visible? Does the structure still make sense in Arabic? Can the same semantic status work across the client and internal products?
The foundation work covered color, typography, grid, spacing, elevation, iconography, logos and state illustration. Color decisions were evaluated for contrast instead of being accepted because they matched the brand. The documented primary-green test, in the palette at the top of this page, showed that lighter green values failed for normal text, while a darker green reached a 5.3:1 contrast ratio and passed the shown normal and large text checks.
That changed the role of the palette. A brand color was not automatically an interface color, so usage had to account for text size, background and interaction context.
Tokens create a bridge between design intent and product implementation. Instead of telling teams to use green 700, semantic names describe the role of a value: text default, text subtle, border danger, background selected or icon disabled. The proposed hierarchy has four steps, shown here with one example.
Token hierarchy, with one example
This naming model lets the underlying value change without forcing designers and developers to reinterpret every screen.
| Early issue | Analysis | Design response |
|---|---|---|
| Repeated product screens used similar but separate patterns | Visual similarity did not guarantee shared behavior | Defined foundations and semantic roles before consolidating components |
| Brand green was used too broadly | Some combinations failed readable contrast | Tested palette roles and used darker values where text contrast required it |
| Components were described by appearance | Default-only components hide interaction gaps | Added state coverage for hover, pressed, focus and disabled behavior |
| Arabic was treated as a translated screen | Direction changes structure and interaction order | Added Arabic variants and evaluated layout, alignment and icon direction |
| Library work stopped at components | Teams still had to reconstruct common workflows | Added product templates for recurring layouts and flows |
| New patterns could be added informally | Uncontrolled growth recreates duplication | Defined request, validation, review, versioning and documentation steps |
Select is a useful test because it combines text, icons, a floating panel, many states and both writing directions. Components were organized by platform, interaction state and theme, so a definition covers more than its default appearance. The state model includes default, hover, active or pressed, keyboard focus, disabled, and loading, empty and error where relevant.
The documentation explains appropriate usage, anatomy, properties, variants, states, semantic tokens and accessibility behavior, instead of presenting only a visual specimen. Platform differences are explicit: shared components keep the same purpose and semantic tokens while adapting layout, target size, hover behavior or navigation pattern for web and mobile.
Accessibility sits next to the component rather than in a separate document. For Select, the documentation covers keyboard behavior, focus, screen reader behavior, ARIA, motion and contrast. The keyboard table below defines what happens on each key press, from opening the panel to closing it and returning focus.
I validated the rules at three levels:
The last level checks whether the same semantic status and spacing rules can support both a guided client flow and a dense internal workspace. This is artifact-level validation, not usability evidence.
Templates capture recurring workflows and composition rules. The template file includes multi-column layouts, login and sign-up, onboarding, dashboard, inbox, tasks, chat, detail views, progress steps, document management, search and filters, notifications, profiles, permissions and help, with English and Arabic variants and, in several areas, desktop and mobile compositions.
Arabic is treated as a structural variant rather than a translation. In the mobile sign-in below, direction, alignment, icon placement and the language control mirror together.
The multi-step form shows the same idea at desktop width. The progress rail, labels, field order and actions all move to the right-hand side in Arabic, while the components and tokens stay the same.
A token or component that works alone can still fail inside a dense layout, so templates also act as a stress test for the system. The analytics template composes breadcrumbs, a page title with tags, an alert, tabs, metric cards and a reusable chart family. The operational template combines summary cards, view tabs, search and a data table.
The SIDF Client Portal and SIDF Internal Portal case studies show the same system language applied to complete client and internal workflows.
A system stays coherent only if teams know how to change it. The proposed contribution flow has six steps:
Contribution workflow in six steps
This turns governance into a service for product teams rather than a gate that only says no.
The presentation shows a token workspace with 264 variables. The supplied template file contains no local variable collections or local styles, so it is the template and application layer, not the full source library. The likely explanation is that the templates use a separate library, but I cannot verify that from the template file alone. I therefore present this as an evolving design system and template library, and I do not claim:
Innonative is an AdTech platform that brings campaign creation, sponsored-ad setup, approvals, audiences, reporting and publisher implementation into one operational product.
As the sole product designer, I led the work from research and workflow definition through interface design, prioritisation, design-system development and engineering handoff. I worked with a product manager, product owner, tester and three developers to turn a technically complex advertising process into connected product journeys.
Creating a native-advertising campaign involves more than uploading an advert. Teams need to configure campaigns, prepare creatives, manage approvals, define audiences, fund accounts, monitor activity, build reports and support implementation on publisher websites.
Each task carries its own information, rules, dependencies and states. When those tasks are designed independently, users have to understand the internal organisation of the platform before they can understand what to do next.
The challenge was to organise this complexity into a coherent product, one that could support the full campaign lifecycle without oversimplifying the operational work behind it.
How might we make a technically complex advertising workflow understandable across creation, approval, reporting and implementation?
I conducted the research and interview work together with the product manager and product owner. We used those conversations to understand the platform requirements, operational tasks and areas that needed to be created or improved.
The original research records are not included in this portfolio, so I do not publish participant numbers, quotations or unsupported findings. The case study instead focuses on the workflow questions and product decisions that are preserved in the final design.
I translated the research and product discussions into a working prioritisation file. Proposed features and enhancements were grouped into must-have and nice-to-have categories so the team could distinguish essential workflow support from improvements that could be delivered later.
This connected design recommendations to product planning. It also gave discussions with product and engineering a concrete structure: what problem the feature addressed, why it mattered and whether it belonged in the immediate scope.
The research made it clear that campaign creation could not be designed in isolation. The product needed a broader structure connecting the work before, during and after campaign setup.
I organised the experience into principal product areas:
This product architecture gave each task a clear home while preserving its relationship to the wider campaign lifecycle.
Campaign creation needed to collect detailed information without presenting the process as one dense configuration screen.
I structured the journey around clear stages and supporting states:
The interface uses sections, progressive steps, summaries and confirmation states to help users understand both what they are configuring and where they are in the process.
The dashboard then changes the user's task from creation to monitoring. Charts, filters, metrics and tables provide different levels of detail without forcing every piece of information into the same visual hierarchy.
Campaign configuration defines the broader activity, while ad creation handles the creative material and sponsored products that appear within it.
I designed the ad-creation experience to support configuration, content selection and preview before the work enters the approval process. This allows users to see how the advert will appear rather than treating the creative as a collection of form fields.
Keeping campaign and ad creation as connected but distinct journeys helped reduce confusion between campaign-level decisions and individual creative decisions.
Much of the platform's complexity comes from work that moves between people and states. Campaigns, creatives, advertisers and reports can be incomplete, awaiting review, approved, rejected or ready for the next action.
I used shared patterns for:
Creative and advertiser approvals were given dedicated spaces rather than being hidden inside campaign setup. This makes review work visible and keeps responsibility clearer for the people involved.
Reporting needed to support both advertiser and publisher perspectives. The product file includes dedicated views as well as report creation, scheduling and saving.
I designed reporting as an operational workflow rather than a static analytics screen. Users can move from selecting the relevant information to configuring, scheduling and reviewing a report.
Tables provide detailed records, while charts and metrics help users understand patterns and performance at a higher level. Shared filters and controls maintain consistency with the campaign dashboards.
The native-ad widget is a useful example of the platform connecting a business task with a technical one.
The flow allows a user to:
The work also covers creative rejection and implementation through a tag-management workflow. This meant the design had to support both visual configuration and a clear handoff to the person responsible for placing the widget on a publisher's website.
The breadth of the platform required more than a collection of reusable buttons. Campaign creation, dashboards, approvals, reporting and account management all use different combinations of forms, navigation, data display and feedback.
I led the design-system work as part of the product, organising it through an atomic structure that moves from shared elements into larger application patterns and page examples.
The system helped the product maintain a consistent interaction language while supporting dense configuration and operational screens. Design annotations also gave the team a clearer way to communicate measurements, notes and expected behaviour during handoff.
I worked mainly with the product manager and product owner to understand requirements and prioritise the scope. The must-have and nice-to-have file helped us discuss proposed capabilities in terms of user and delivery value rather than treating every request as equally urgent.
I collaborated with three developers to check feasibility, clarify interaction behaviour and prepare the designs for implementation. A tester supported the delivery process by reviewing the implemented experience and identifying issues that needed design or development attention.
Many flows in the Figma file are marked Ready for dev, reflecting the structured handoff state of the work. This case study does not claim that every designed area was released or that the product produced unverified business results.
The product manager, product owner and I conducted research and interviews to understand advertising tasks, requirements and workflow problems.
Output: workflow questions, product requirements and areas for creation or improvement.
I translated the research into proposed features and enhancements, grouped as must-have and nice-to-have.
Output: a prioritised design and product scope.
I organised the platform around the campaign lifecycle and the responsibilities surrounding it.
Output: product architecture covering campaigns, ads, approvals, audiences, reports, accounts, users and widgets.
I mapped and designed the principal journeys, including campaign creation, ad setup, approvals, reporting and widget implementation.
Output: connected flows, interaction states and high-fidelity screens.
I created and applied shared foundations, controls and application patterns using an atomic structure.
Output: a design system supporting forms, dashboards, data views and feedback states.
I worked with the developers and tester to clarify behaviour, prepare designs for development and review implementation issues.
Output: development-ready flows and clearer design guidance.
The final work defines a broad AdTech product covering the operational journey from campaign and ad creation through approvals, audiences, reporting and publisher implementation.
It is supported by a design system that connects foundations, shared controls and application-level patterns across complex forms, dashboards and data-heavy workflows.
The case study does not claim unsupported improvements in setup time, abandonment, adoption, revenue or campaign performance.
Innonative strengthened my ability to design across an entire operational product rather than treating each feature as an isolated request.
The most important decision was to connect the surrounding work (approvals, audiences, reporting, account operations and implementation) to the campaign lifecycle. The design system then gave those different areas a common interaction language.
If I revisited the project, I would preserve the research and decision history more systematically and define measurable experience baselines before development. That would make it easier to compare the final product with the original workflow and evaluate where the design created the greatest value.
Jardin Bio explores an alternative to the standard online-supermarket experience. Instead of asking customers to begin with a large catalogue, the concept introduces weekly boxes built around seasonal produce, while still allowing people to create and save a box that fits their preferences.
I developed the idea into a responsive commerce experience covering discovery, box creation, saved selections, wishlist, bag, checkout, accounts and editorial content. I also created the visual foundations, reusable components and a sticker system that carries the brand into the delivered box. The homepage is published on Dribbble.
Buying groceries online usually starts with categories, filters and long product lists. That structure works for a large catalogue, but it does little to communicate what makes seasonal organic produce different.
Jardin Bio needed its own way of shopping, one that could help customers choose fresh produce without building every order from scratch, while still giving them control over what they receive.
The experience also needed to support the commercial reality of fresh produce: availability changes with the season. Rather than hiding that constraint, I used it as the organising idea for the product.
How might we make seasonal produce easier to choose without turning the experience into another fixed box or conventional supermarket catalogue?
I began by exploring how people approach buying fresh and organic produce, what they expect from online grocery shopping and how seasonality affects the products a business can offer.
The discovery work included market and experience research alongside exploratory interviews. The original notes are no longer available, so I cannot responsibly publish participant numbers, direct quotations or claim that a specific behaviour was validated. Instead, this case study focuses on the product questions that shaped the work and the decisions preserved in the final design.
The surviving design work and project rationale reflect four themes. These are presented as a retrospective synthesis rather than quoted research findings.
A large catalogue provides flexibility, but it also asks customers to make many individual decisions. This led me to explore a curated starting point rather than making product-by-product selection the only path.
Prepared boxes simplify the experience, but personal preferences still matter. This created the need for a second path where customers could control the contents.
Seasonal availability is fundamental to fresh produce. It became part of the product structure through weekly selections instead of remaining hidden as an operational constraint.
The physical box is the moment when the promise of freshness becomes real. This encouraged me to connect the interface and delivery experience through a family of produce stickers.
I translated these themes into a product model with two connected paths:
This became the foundation for the information architecture, user flows and interface system.
Before designing detailed screens, I organised the product around the main customer tasks:
This structure helped separate the core box journey from the supporting commerce and account experiences while keeping them connected through consistent navigation and patterns.
I mapped the experience around three recurring journeys:
These flows established what information and actions needed to remain visible at each stage before I moved into detailed interface design.
I used the flows to explore how the box should behave as a product object, not merely as a shopping cart with a different name.
The key interaction questions included:
I developed the experience from page structures and task flows into responsive high-fidelity screens. The design was refined across the home experience, product categories, box creation, saved boxes, bag, checkout and account areas so the central idea remained coherent throughout the product.
The surviving file documents the final direction rather than every discarded alternative. For that reason, the case study explains the decisions visible in the final experience without claiming unsupported test results.
I designed two complementary paths instead of forcing every customer through the same shopping behaviour.
Weekly boxes give customers a prepared starting point based on produce available in season. This reduces the effort of selecting every item and gives Jardin Bio a natural way to feature what is fresh now.
Customers who want more control can build their own box from the available produce. They can review the selection as one collection, save it and return to it later.
These are not separate features. Together, they balance three needs:
In a typical commerce experience, the cart appears near the end. In Jardin Bio, the box becomes visible much earlier and acts as the organising object for the whole journey.
The flow moves through five clear moments:
This makes the experience feel less like collecting unrelated items and more like preparing something personal.
The custom-box flow needed to provide freedom without losing structure. Customers can browse produce and build a selection while keeping the box itself visible as the main object they are creating.
I designed the experience to answer three questions throughout the flow:
Product cards, selection states and the box summary use consistent patterns so that customers can move between browsing and configuring without learning a new interaction model.
Saving a box extends the idea beyond one purchase. A preferred combination becomes something customers can find again, rather than rebuilding it from the beginning each time.
The experience does not end at checkout. I created a family of produce stickers to accompany the delivered box and connect the digital product with the physical moment of receiving it.
The stickers give individual products a recognisable Jardin Bio detail and make the box feel considered rather than anonymous. They also give the identity room to be warmer and more expressive than the functional parts of the interface.
This became an important part of the concept: the box is both the product model inside the website and a physical brand touchpoint after delivery.
The concept extends beyond the seasonal-box idea into the supporting journeys needed for a coherent commerce product.
I designed:
The editorial experience gives Jardin Bio space to talk about produce, seasonality and food beyond individual product listings. This supports a brand that wants to guide and inform, not only sell.
The product combines expressive imagery and colour with repeated commerce interactions. To keep those two sides coherent, I created foundations and reusable patterns before applying them across the full experience.
The system includes:
Product cards needed enough flexibility to work in category pages, recommendations, wishlist, box creation and smaller mobile layouts without losing consistency. The shared foundations also helped the editorial and account experiences feel like parts of the same product.
I explored organic-shopping expectations, seasonal produce and comparable commerce experiences, supported by exploratory interviews.
Output: questions, opportunity areas and an initial product direction.
I framed the tension between the convenience of a prepared box and the control of selecting individual products.
Output: a two-path model combining weekly seasonal boxes with custom, saved boxes.
I organised the information architecture and mapped the main journeys across discovery, box creation, saving, bag, checkout and account management.
Output: experience architecture and task flows.
I developed page structures and interface directions for the core journeys, considering what customers needed to see and do at each step.
Output: screen structures and evolving product flows.
I defined the visual foundations and reusable components, including product-card families, controls, fields and sticker elements.
Output: a reusable interface system connecting commerce, editorial and account experiences.
I applied and refined the experience across desktop and mobile, adjusting hierarchy and actions for each context.
Output: responsive high-fidelity product screens covering the principal journeys.
I reviewed the complete experience for continuity between seasonal discovery, customisation, saving and checkout. Because the original evaluation records are unavailable, I do not claim formal usability results.
Output: a complete product concept and a clear set of questions that should be validated before development.
I designed the principal journeys for both desktop and mobile rather than treating mobile as a reduced version of the larger screen.
On desktop, the layout can show the catalogue and box context together. On mobile, the same information has to be prioritised more carefully, with selection, summary and next actions remaining clear in a narrower space.
The responsive work covers the home experience, categories, wishlist, bag, box creation, checkout, account, profile and editorial pages.
The final concept defines a complete responsive commerce experience built around a distinct product idea. Seasonal boxes provide guidance, custom and saved boxes preserve personal choice, and the sticker system carries the brand into the physical delivery.
The work covers the journey from discovery to checkout and is supported by reusable foundations and components across desktop and mobile.
No launch or performance results are claimed for this concept.
The strongest part of Jardin Bio is the relationship between the business model and the customer experience.
Seasonal availability is not treated as a limitation hidden behind a catalogue. It shapes the way products are introduced. Curated boxes reduce the effort of choosing, custom boxes maintain control, saved boxes support returning customers, and the stickers connect the interface to the delivered product.
Looking back, I would document the early research and decisions more rigorously and validate the box model with customers before development. The concept nevertheless demonstrates how product structure, interaction design, visual identity and physical details can work together around one clear idea.
Before moving toward development, I would test: