Getting ideas across

RialFit

A matching engine serving two businesses at once: a white-label product for each developer, and a broker channel where we capture leads and lend them out. The proposed data model treated ownership, management and origin as one thing. I tested the product myself, rejected the model in writing with three contradictions and six edge cases, and redefined the domain around three relationships. The change shrank from over 3,000 lines to under 900, and the decision meeting took 34 minutes and three yes-or-no questions.

Multichannel model, attribution and metrics in production one month after the spec. Now in validation with 38 developers in the catalog.

1monthSpec to production−70%Size of the change

RialAds

The marketing agency inside the same unit, on the business strategy side. I set up how it operates: roles, a reporting process, a KPI glossary, and a dashboard with its success criteria written before anyone built it.

Fidi

Virtual accounts, closed wallets, cash pooling and cards that retailers and fintechs launch under their own brand, on top of banks that are already integrated. I led the experience and started by defining what the product is before what it looks like: an operational tool, not a marketing surface. I built the design system from business states outward, so that active, blocked, pending and operating mean one thing everywhere, then packaged it so engineering and AI tools could build from it without me in the room.

A design system engineering and AI tools build from without me, and a white-label wallet flow, built on the same stack in a day, now used to sell the product to retail clients.

4yrsLeading the experience1dayTo a consumer wallet

beClassmates

I started where students are: a redesigned rewards module, prototyped by me and shipped by engineering, and a community reframed from blog to plaza once I saw its silence was psychological, not technical. When the draws failed, I turned the rules into a test plan before anyone tested. For teachers, an hours log where attendance is the only truth. Then the operation itself: I sat with the people running enrollment and payroll by hand and scoped by pain, 80% of it, not 100%, in ten specs with an explicit “not included” list.

Rewards in production since May. Operations backoffice specified and prioritized in four weeks, first sprint scoped to moving 500 people onto the new system.

4weeksTo a prioritized backoffice10Specs engineering builds from

Maré

I parsed 9,000 messages from two clubs: 83% were copies of the list, and 30% of cycles lost confirmations to concurrency. I corrected my own occupancy number when it didn’t match the club’s prices, sized the market with a manual census, and designed a link that lives in the group. Rules: the coach approves, never drafts; nothing about the sea gets automated. Priced, with a gated pilot and two different decks for the club and the partner.

Pre-pilot. Brand, prototypes, and a plan that can stop without debt.

9,000Messages parsed30%Of cycles lost confirmations