Minerva
Role
Sole UI/UX Designer
Scope
Mobile, Desktop Web App, Conversational AI UX
Duration
July 2025 – October 2025
Overview
I joined a fast-paced 2-person engineering team to elevate an existing mobile prototype into a production-ready, cross-platform product. Over a sprint, I led the end-to-end design for an AI Tutor conversational chatbot, built a unified light/dark design system across mobile and desktop, and designed monetisation and setting flows.
The Challenge
Joining an existing codebase with pre-rendered mobile screens presented a unique set of constraints that required strategic, systematic design execution:
1.
Reverse-Engineering Mobile Logic
Analyzed legacy high-fidelity screens without existing documentation to map hidden dependencies, screen-by-screen flows, and core interaction patterns.
2.
Complex Conversational UX for AI
Designing the "AI Tutor" required translating traditional test, practice, and study planner features into a fluid, prompt-driven chatbot interface.
3.
Evolving Backend Constraints
Defining unique AI "Study Modes" required close collaboration with developers to balance user-facing clarity with backend prompt engineering.
The Design Direction
We aimed to establish a scalable multi-platform design foundation, optimize conversion copy for premium subscriptions, and craft an intuitive AI chatbot interface with guided, action-based response loops.
The Solution: Core Features
SOLUTION 1
Conversational AI Tutor & Custom "Study Modes"
Custom chart frameworks transform dense medical metrics into readable, actionable insights.
SOLUTION 2
Responsive Desktop & Multi-Pane Architecture
Adaptable UI patterns accommodated shifting backend capabilities without breaking the user experience.
SOLUTION 2
Clear, value-driven microcopy turns complex account settings and pricing tiers into confident user decisions.
Adaptable UI patterns accommodated shifting backend capabilities without breaking the user experience.
System Architecture & Handoff
To ensure long-term scalability across future roadmap items (such as upcoming gamification and leaderboards), I retrofitted existing screens into a unified, component-driven design system. I tokenized typography, iconography, and multi-state variants across both light and dark modes, giving the engineering team a clean, reusable component library that streamlined sprint handoffs.
The Impact
Establishing a cohesive design framework accelerated development and set the stage for platform growth.
Multi-Platform Readiness: Successfully scaled a mobile-only prototype into a responsive desktop web application ahead of critical deployment deadlines.
Reduced Friction in AI Learning: Action-based prompts and distinct study modes eliminated dead-ends in chatbot interactions, keeping students engaged in active recall.
Learning Outcomes
Effective AI design requires balancing open-ended conversational freedom with guided UI guardrails.
Don't rely solely on a blank text box; use suggested prompt chips, action buttons, and clear mode selectors to guide users toward valuable outcomes.
When joining a project mid-flight, retrofitting existing screens into a structured component library pays off immediately when scaling to new platforms like desktop.