Solo-designed on top of a product three agencies had already touched. Extended it with new features, rebuilt the parts that were quietly broken, and gave it a real responsive system instead of a shrunken web layout. Four months, design delivered.
Saudi ArabiaAmplifidor is an influencer platform for Saudi Arabia and the wider region, built to give influencers, brands, and agencies one place to run their work. By the time I came in through BitBang, the product had passed through three agencies. It had features, but it didn't read as one product. My job was to keep extending it while fixing the parts that were already wrong. The hardest problem wasn't adding screens. It was making a product built by four different hands feel like it was built by one.
The platform brings influencer profiles, cross-platform analytics, an affiliate engine, a two-way wallet, and community groups into a single account. Brands use it to find and connect with the right talent in under an hour instead of scouting cold. Influencers use it to hold their whole presence in one place. The premise is consolidation, so the product itself had to feel consolidated. That was exactly what three rounds of separate agency work had eroded.
Influencers and brands were being handed a product that looked finished and behaved like three products stitched together.
Discovery was light by necessity. This was a contracted build on a product with two years of prior work behind it, not a greenfield start, so my first weeks went to reading what existed and finding where it broke rather than running a research phase I'd be inventing. Working next to the PM, CTO, and the dev teams gave me the real constraint map: what was cheap to change, what was load-bearing, what I'd have to design around.
Good meant coherence, not a metric. Could a new person move between the dashboard, a profile, analytics, and the agency tools without feeling the seams? Could the same flow work on a phone because it was designed for one, not resized into one? I didn't own analytics on this project and it didn't reach the point where launch numbers existed, so success was judged on the design holding together across every surface I touched, and on engineering being able to build from it without guessing.
The responsive system was faking it. What shipped as "responsive" was the web layout scaled down. It technically fit a phone and worked for nothing. I treated small screens as their own design problem, not a resize target, and built a new component set and layout logic for them. This was the clearest inherited failure on the project: something that had been called done, observed to fail in the hand, diagnosed, and rebuilt properly.
One account, many identities. A single login could hold several identities, personal and agency, each with a different set of things you could do and see. That meant a profile switcher that changed not just the name at the top but the whole capability underneath it, personal settings and rates in one context, admin and agency controls in another. The architectural call was keeping all of this inside one account with a clean switch, rather than splitting it into separate logins that would have fractured the person's presence, which is the one thing the product exists to unify.
Fixing components without breaking the build. Some of the design system had genuinely wrong UX baked in, and engineering had already built against it. I couldn't just redraw it. Each correction was a judgment about whether the fix was worth the rework it forced downstream, and I made those calls in conversation with the devs rather than over the wall.
Onboarding and the front door. New sign-up, sign-in, and verification flows, plus onboarding that actually set someone up rather than dropping them into a blank account. The first screens were among the most inconsistent inherited pieces, so they were worth rebuilding outright.
The corrected design system is the connective tissue. It's what lets the parts three agencies never coordinated read, finally, as one product.
Directionally: the foundation was solid, the seams were closing, and the product finally had one voice across its surfaces.