Redesigning Scalable Mentorship Platform

Redesigning Mentorship Platform cover image

Thesis: A five-year legacy application with feature bloat needed a design system foundation before iterative improvements could scale.

Outcomes: 50+ reusable components (70% dev time reduction) • DAU: 32% → 47% • 40% increase in successful mentor-mentee matches • 10,000+ messages/month • Contributed to Ruby implementation

Before → After: Feature-bloated legacy app → Systematic, scalable design • Inconsistent UI → Unified design system • 32% DAU → 47% DAU • Manual matching → Smart algorithm-driven matching

The Challenge Stellarup secured €500K in seed funding with a clear mandate: improve UX and drive sales. The product—a mentorship platform connecting professionals—was functionally sound but, after five years of unstructured feature additions, it was bloated and inconsistent.

What I inherited: Five-year legacy application with no design system

Feature bloat masking core value

32% DAU despite 5,000+ users

Manual matching that wasn't scalable

The brief: Transform UX, establish design foundations, and improve engagement.

I spent two months conducting a comprehensive UX audit—reviewing every feature, mapping flows, and identifying technical debt.

The critical insight: Before fixing individual features, the product required a design system foundation. Without it, every improvement would compound existing debt.

I partnered with Julian Roux-Stevens (CPTO) to build a phased roadmap:

Design System → Foundation for scale

UX Refinements → Fix core features

Smart Features → Matching algorithm and data visualisation

Building the Design System I led design system development in Figma and contributed to Ruby implementation using ViewComponent (GitHub's architecture framework), Lookbook (component previews), and Stimulus JS (interactivity).

dESIGN_SYSTEM.png 200 KB

Key Deliverables: Foundation: Colours, typography hierarchy, 4px grid spacing, icons, illustrations

80+ Components: Buttons (with interaction states), forms (inputs, validation), cards (content, profile, tooltips), sortable tables, navigation, data visualisation

Documentation: Voice, tone, layout principles in Confluence • WCAG 2.1 AA accessibility standards

Impact: 70% reduction in development time for new features, unified UI, scalable foundation.

UX Refinements & Smart Features With the system in place, I focused on iterative, data-informed improvements.

Core UX Fixes

Before vs After - dashboard.png 1.61 MB

Before vs After - Page Header.png 368 KB

Dashboard: Reduced sidebar items using contextual tabs, added tooltips to Insight cards

Programme Management: Streamlined CRUD flows, tile/list views, sortable tables

Smart Matching Algorithm I designed the UX for a collaborative filtering system that analysed user profiles (skills, goals, industry, availability) and suggested optimal mentor-mentee pairings with match scores.

How it worked: The algorithm analysed profile compatibility and surfaced high-percentage matches with clear reasoning (e.g., "92% match: Both interested in AI product design").

10,000+ messages/month across 5,000+ users

Data Visualisation I redesigned the Insights dashboard to show:

Real-time messaging metrics (response rates, active conversations)

Usage patterns (peak activity times, active segments)

Forecast vs. actual engagement

Impact: DAU improved from 32% to 47%—a 15-point increase achieved through behavioural analytics and A/B testing.

Pages combined.png 2.82 MB

Impact

Metric | Before | After | Delta Daily Active Users (DAU) | 32% | 47% | +15 points Component Reusability | 0 | 50+ | 70% dev time reduction Successful Matches | Manual/Low | Algorithm-driven | 40% increase Monthly Messaging | N/A | 10,000+ messages/month | High Engagement

What I Learnt

Systems > Screens The 70% dev time reduction proved design systems are infrastructure, not cosmetics. Building foundations first enabled everything that followed.

Audit deeply before acting Two months of understanding legacy pain points prevented building on broken foundations. Strategic patience pays off.

Code + Design = Force Multiplier Contributing to Ruby implementation let me bridge design and engineering, shipping faster with fewer handoff errors.

Metrics validate, intuition guides DAU improving from 32% to 47% was the result of data-informed iteration. But building the design system first? That was conviction.

Role: Product Designer Team: Julian Roux-Stevens (CPTO), Product & Engineering Team Timeline: October 2023 – present Impact: 50+ components • 70% dev time reduction • DAU 32% → 47% • 40% match success • 10K+ messages/month