SpeakUp: AI Interview Coach
Vanessa Chang
Transforming AI Logic into a Supportive Coaching App
I translated complex LLM outputs into a high-fidelity, functional interview coach, focusing on "Moment-level Feedback" to help non-native speakers build real-time communication confidence. My approach prioritized making AI critique feel like a supportive mentor, not a judge.
Translating raw AI analysis into actionable metrics—Confidence, Clarity, and Pacing—to reduce user cognitive load
The "AI-First" Approach
Unlike traditional projects, I didn't just stay in Figma. I utilized an AI-assisted development workflow to validate how AI feedback should feel. This hybrid approach allowed me to lead the end-to-end creation of an AI-driven product in weeks, not months, bridging the gap between design vision and functional reality.
Key Values Delivered
I moved beyond visual fixes to create a systemic framework for the team:
Translating AI Complexity: Designed the information architecture to turn raw LLM scores into a "Qualitative Delivery Analysis" (Confidence, Clarity, Pacing) that users can actually act on.
Strategic Pivot to "The Moment": Through rapid testing, I designed the "Moment-level Feedback" UI to provide precise insights, showing exactly what happened at specific timestamps in their speech.
Rapid MVP Delivery: My approach demonstrates how a designer can lead the end-to-end creation of an AI-driven product in weeks, not months.
The Impact
A functional prototype that proved "specific coaching" beats "generic summaries." 100% of user testers felt more prepared for real-world interviews because the feedback was anchored to their actual performance timestamps.
Moment-level feedback logic: Mapping AI critiques to specific timestamps to build user muscle memory and speaking confidence
For a deep dive into my AI-UX framework: View the full case study on my portfolio
AI UX Design · Rapid Prototyping · Information Architecture · Product Design · Google AI StudioLike this project
Posted Feb 1, 2026
Designed & prototyped an AI coach for non-native speakers. Transformed LLM data into actionable, moment-level feedback using an AI-assisted development workflow
Likes
0
Views
1
Timeline
Nov 20, 2025 - Dec 4, 2025
