Frego connects buyers and suppliers in one place: product sourcing, requests for quotes, communication and logistics. The product was already live and earning money. The goal was not to rebuild it from scratch, but to improve usability, clarity and trust within existing business and technical constraints.

The company ran ads to the landing page, and traffic was strong. But only 10% of people who started registration finished it. Paid traffic was leaking at the very first step.

The cause was in the flow itself. People had to go through three steps (basic info, account type and a full company profile) before seeing anything useful. We asked for company details before giving anything back.

I split the information into two groups: what is needed to create an account, and what can wait until the person is inside the product.

• Registration went from 3 steps to 1 screen. Name, email, country and role. Role selection became cards that explain what each account type gets, so people choose with confidence.
• The company profile moved into the product. Instead of a separate onboarding flow, first-time guidance became a 5-step checklist on the dashboard: complete the profile, verify the company, add the first product, review requests, start a conversation. Each step says why it matters, for example that verified suppliers get more responses.

I designed the supplier dashboard around the platform's core actions: active products, new requests (with unanswered ones highlighted), closed deals and live shipment tracking. Suppliers see what needs attention without digging through sections.

A plain chat didn't build trust in B2B deals. The production version added business context to every conversation (product, request, request type), a company and manager panel with verification status, filters for active dialogs and files inside the chat. For this case study, I redesigned its interface on top of that logic.

A component library for a data-heavy B2B product, plus multi-level navigation across the header, sidebar, nested menus and company settings.

Track checklist completion to see which onboarding step loses people, and test whether verification prompts actually raise supplier response rates.

Nodera is a partner-program platform: partners earn commissions from direct clients and from their network. For them, the wallet is the main point of trust. If money feels unclear, people stop trusting the whole product and start writing to support.

1. How much can I withdraw right now?
2. Why can't I withdraw the rest, and when will it unlock?
3. Where is my payout?
4. What happens if something goes wrong?

Every decision below answers one of them.

Available, on hold, in transit and paid are shown separately, so the main number is always the one you can actually withdraw. Held commissions are listed with dates and sources, plus a plain reason: commissions wait 30 days in case a client asks for a refund.

Every payout has a timeline: requested, processed, sent to the bank, arrived. When a payout fails, the screen says why (the bank account was closed), confirms the money is back in the available balance, and offers the next step: send it to another account or remove the closed one.

1. AmountAvailable balance and minimum shown right under the field
2. AccountClosed accounts are disabled and marked, so a failed transfer can't happen twice
3. ReviewFee, the amount you receive and the arrival date before you confirm
4. PINOne last check that names the amount and the account

The Security page shows when the product asks for a PIN, two-factor authentication or identity verification, and why. Extra steps feel protective rather than random. Each 2FA method states its trade-off, for example that email is the least secure.

Boost Solutions uses AI across departments, from sales analytics to automated customer replies. The existing dashboard was meant to show business performance and AI's impact on it, but it did neither clearly.

The brief: analyze competitors, audit the existing wireframe, create a moodboard and a design concept.

I reviewed four AI-focused BI platforms: ThoughtSpot, Power BI, Domo and Databricks One. They share the same weaknesses: overloaded screens, a confusing hierarchy, rigid filters and AI insights that don't explain themselves.

That set the direction: clean navigation, detailed KPIs through interactive charts, and AI insights that come with explanations and forecasts.

• No filters in key sections. People couldn't change the period or analyze trends.
• An unclear AI block. Percentages by department with no explanation of what AI actually changed.
• No drill-down. Sections weren't clickable and didn't open details.

• Period filters on every KPI card, so each metric can be explored on its own.
• AI impact by department as explainable cards. Each shows the change, a trend and a plain-language reason, for example that AI lead scoring raised conversion.
• Before and after AI in sales and operating costs, so AI's value is shown in numbers, not just claimed.
• Clickable cards that open detailed views.

1. Analysis5 h 40 min: competitors and UX audit
2. Wireframing4 h 20 min: sketches, structure, prototype
3. Moodboard3 h: type, color, charts
4. UI design7 h 30 min: components, charts, screens
5. Presentation4 h 10 min

This was a test assignment, so the design wasn't tested with users. It's built to cut the time needed to find key metrics and to make AI's contribution clear to both management and operations teams.