← AbdelRahman Atif

Case study · Moneyfellows

Designing for credit risk

Leading design on credit assessment — an ML model built with the data science team, and the experience that made it work.

Behance ↗
Role
Senior Product Designer
Company
Moneyfellows — fintech
Timeline
2023 — 2024
Partners
Data science, engineering, stakeholders across risk & product

−12%

default rates

56M EGP

decrease in exposure

+10%

retention lift

Context

Moneyfellows digitizes money circles — the rotating savings groups (ROSCAs) that much of the region already runs on trust. Digitizing trust means underwriting it: when a member receives their payout early in the circle, the company is carrying real credit risk. Getting that assessment right is not a feature; it's the business.

The project

The credit assessment work was driven by a Central Bank of Egypt licensing requirement, which put it at the center of the company's regulatory and business agenda at once. I led design on it, working directly with the data science team implementing the ML risk model, and owning the side of the system users actually touch: how credit limits are communicated, earned, and understood.

The design answer was gamification with real stakes. We turned creditworthiness into a coins mechanic: members build their limit through behavior the model can trust, and the interface makes the path to a higher limit legible instead of opaque. A risk model only works if the product around it collects honest signals and communicates decisions people can accept. I sat with the CEO and stakeholders throughout to keep the mechanic, the model, and the business case aligned.

Three Fellows Coins app screens — onboarding with first coins, slot selection with coin fees, and the My Vault coin ledger
The mechanic end to end: earn your first coins, spend them on a payout slot, track every movement in the vault.
Slot picker showing months with coin costs, fees, and cashback — filterable by fastest, no fees, or high return
The core trade-off made legible: earlier payout slots cost more coins. Filters let members optimize for speed, fees, or return.
The Fellows Coins currency — golden coin designs with the brand mark
The currency itself.
Three coin states — green for earned, red for deducted, blue for used
Three states, three colors: earned, deducted, used.
Fellows Coins asset system — coins, bilingual logos, fee cards and color variants
The asset system, bilingual from the start — including the Arabic wordmark.
Campaign poster — simplified approach to financial literacy
Financial literacy campaign, English.
Arabic campaign poster — a simpler way to raise financial awareness
And in Arabic — same voice, native register, not a translation.
My Vault screen on a phone held in hand — the coin ledger showing earned, deducted and used entries
My Vault: the ledger that makes the credit score tangible — every payment, document, and circle joins the record.

How the teams worked

Cross-functional projects fail in the seams, so I spent deliberate effort on them. I facilitated workshops that mentored designers, aligned engineers around the Figma workflow and design system, and walked stakeholders through new experiences before they shipped. The same period included rebuilding the design system's usability — designer satisfaction rose by 60% — which made every downstream collaboration cheaper.

Outcomes

The credit assessment shipped with the ML model in production: default rates dropped by 12%, exposure decreased by 56M EGP, and retention lifted by 10%. It remains the clearest proof point in my work that design earns influence by moving the numbers the business actually runs on.