BRENDAN BROWN

My independent product studio. Three iOS apps live on the App Store in a year, two more in development.

• Built all of Cardy: iOS and web apps, marketing and admin sites, social automation, a teleprompter app.
• 30% download-to-paid on Cardy, by rebuilding onboarding as a guided proof of the product's own reasoning rather than a feature tour.
• 30% download-to-paid on Cardy. The call: a reading that feels like an assertion is easy to dismiss, so onboarding became a guided proof (birthday in, cycle runs, card out) instead of a feature tour. That rate is what made paid acquisition worth buying, so Apple Search Ads went against the category's own vocabulary.
• Kept Cardy's core as arithmetic, not a model, because checkable working is the one thing it sells; models only where language is the output, with prompts versioned in the database so wording ships without a build and every reading traces to its prompt.
• Shipped an App Clip because the best demo is ten seconds long: a link or event code gives someone their reading on the spot, and offers the full app only after.
• Built Sniffit and Who2see solo, end-to-end in SwiftUI, Expo / React Native, Next.js API routes, Supabase and RevenueCat, and shipped both to the App Store. Sniffit helped earn $4K+ in flipping profit in six months.
• Sniffit's scan answers in about a second from an on-device Core ML pass, then resolves against Claude Vision and a Turso edge-SQLite comparables query running in parallel, so the screen is never empty while the slow work finishes. The vision call is streamed and the comparables query fires on the identity fields mid-generation, hiding the database round trip inside the rest of the model's response.
• Built the eval harness Sniffit's pricing is tuned against: coverage, precision and recall of the match cutoff against hand-labelled comps, and whether the quoted range contained the real sale price, with saved baselines, so a scorer change is a measured before-and-after. Valuations are graded comp-anchored, blended or model-only by how much market evidence backs them, and the app says which.
• Chose Who2see's poster-extraction model on data, not preference: ran Gemini in shadow behind GPT-4o on live traffic under a byte-identical prompt, logged Jaccard overlap and per-call token cost, and set the density threshold where the cheaper model stops holding up. Prompts in Cardy are versioned rows in Postgres, with the version id stored on every generation, so wording ships without a build and any output is traceable.
• Shipped Cardy after six months in App Review, which rejects its whole category on sight: repositioning what the app claimed to be, written appeals, and eventually a call with the review team.
• Sole design and engineering resource across all three: design systems, native and web architecture, store submission and review, and subscription billing.

NIL marketing platform for brands and college athletes. Led design across brand tools, athlete tools and internal operations, and led the team with direct reports. About $150K of brand deals ran through the platform.

Two-sided NIL marketplace: brands on one side, college athletes and creators on the other. Led product design across brand campaigns, athlete tools and the internal operations behind managed campaigns, and led the team with direct reports. About $150K of brand deals ran through the platform.

• Demand side. 5–10% reply rate on cold brand outreach, from templates that embedded each athlete's own analytics into the email and kept the thread where a reply would be seen.
• Team lead. Led the team with a couple of direct reports: planning, development, QA and show-and-tell; coached a user researcher through usability testing.
• Trust on both sides. A verified brand directory with audience-overlap insight for athletes, and public athlete profiles a brand could evaluate and hire from, so neither side had to act on a claim.
• Marketplace operations. Brands chose self-serve challenges posted to the athlete network or fully managed campaigns run by our team, and brands, athletes and staff tracked deliverables, schedules and earnings in one dashboard.
• Learned where no-code tops out: Bubble got the MVP out fast, then slowed every custom dashboard, interactive game and live payout rule. The lesson: pick tools that can grow with the product.

Home search across responsive web and native app: search, map, listings, credit, and iOS home-screen widgets.

• Designed the user-facing side of a Vantage → FICO score migration: a number people had watched for a year moving for no reason they caused. Pre-launch and launch comms, in-app explainer, dashboard redesign, and a five-state gauge system, designed in full before the switch was paused.
• Designed the home affordability calculator into the whole product: your range overlays the price filter in search, and every home shows where it sits on your scale.

• Pre-sign-in rebuild, 2019. Redesigned with PMs and content strategy for the Information Seeker: first-time buyers, 80% of new accounts. Purchase sales page +16% visit-to-application-start (34.2% → 39.8%), at the largest mortgage lender in the US; refinance +7.5% (20% → 21.5%); homepage +19%.
• 52% vs 48% conversion to credit from the Info Seeker path, which asked where someone was in the buying process and returned a resource dashboard instead of an application form: leads without demanding an application up front.
• Designed asset upload in the application, part of a multi-team effort with bankers coaching clients through adding assets: purchase allocations with an approval letter +37% (19% → 26%); clients clicking to be approved after seeing a solution +46% (48% → 70%).
• Helped originate asking for App Store reviews after positive moments in the app: iOS rating 3.2 → 4.8 stars, visit-to-download +50% (48% → 72%).
• Rocket Rebrand, 2018–19. Four sibling brands rebranded at once alongside Lippincott and Metalabs. Built the rebranded Sketch component library, shipped it with the Spark design-system team, and applied it across responsive web, iOS and Android with several development teams.

• Introduced a user-centered design process to a team that had none, on an Agile Scrum team working directly with stakeholders and users.
• Designed an internal team-member development application, and a web tool that moved thousands of employees between Detroit parking garages.