PrepTalk — AI Interviewer

🏆 Finalist · 1st Place (Education Track)

Recognized for impact on equitable interview preparation and practical, end-to-end execution.

See it in action

The challenge

Interview prep is stressful and time-consuming. Candidates lack realistic practice, tailored questions, and actionable feedback in a single place—especially without mentors or mock-interview access.

Our solution PrepTalk analyzes a résumé, generates role/company-aware questions, and runs mock interviews (HR/Manager/Technical) with webcam and technical coding rounds—then returns structured feedback.

Research insights

Candidates spend hours searching for "likely questions" yet get little personalized practice.
Students want concise company context (culture, values) before mock interviews.
Immediate, specific feedback (structure, tone, confidence) drives improvement.

Design process

Interview Start → upload résumé, choose role/company, select round type
Prep Page → résumé feedback, company snapshot, practice tips
Interview → HR/Manager rounds or Technical editor with prompts
Minimal layout with clear timers and prompts

UI Priorities

High-contrast controls and large touch targets. Panelized feedback with examples and next steps.

Development

Architecture

React Native (Expo) front-end, Flask API, Auth0 auth, Expo Router navigation.

Key work

Structured prompts for reliable JSON (questions, rubrics)
Webcam integration for live mock sessions
Monaco editor + local JS runner for technical tasks
CORS + endpoint testing via Postman before wiring UI

Key features

Résumé-aware Questions: Generates HR/Manager/Technical prompts grounded in your actual experience and target role
Company Prep: Concise summaries of culture, values, and example interview themes to prime candidates
Mock Interviews: Webcam + timed sessions; structured rubrics for tone, clarity, and confidence
Technical Round: Monaco editor with sample tests; JS local runner (extensible to TS/Python via backend)

What's next

User accounts and progress history (PostgreSQL/Firebase)
Realtime voice & posture feedback (WebRTC + TF.js / Realtime API)
Confidence scoring and improved analytics
Public deployment (Vercel + Render)

← Back to Projects View code