get prepped — AI-Powered Interview Training Platform
get prepped is an intelligent AI interview simulation and coaching platform. It helps job seekers practice realistic live technical and behavioral interviews with tailored AI interviewers, dynamic follow-up questioning, real-time speech evaluation, and curated role matches.
Features
- Interactive AI Interview Simulation: Real-time audio and text-based mock interviews powered by Google Gemini and ElevenLabs/YarnGPT/Azure text-to-speech.
- Realtime Voice Interviews: Speech-to-speech interviews over Azure Voice Live, through a WebSocket relay that keeps the Azure key server-side.
- Dynamic Probing Engine: Adapts questioning in real-time based on candidate responses, STAR method criteria, and technical depth.
- Personalized Coaching & Feedback: In-depth post-interview performance breakdowns analyzing vocabulary, technical depth, clarity, filler words, and actionable improvement tips.
- Multi-Region Voice Support: Realistic interview personas with natural voice synthesis supporting US, UK, and Nigerian accents.
- Curated Job Matching & Discovery: Explore vetted tech roles from top tech companies and YC startups matched to your career trajectory.
Tech Stack
- Framework: Next.js (App Router, Turbopack)
- Language: TypeScript, React 19
- Database: MongoDB (Mongoose) for audio caching & user analytics
- Storage: Cloudflare R2 (S3-compatible persistent audio storage)
- AI / LLM: Google Gemini API (
@google/genai / REST)
- Text-to-Speech: ElevenLabs, YarnGPT & Azure Speech
- Realtime voice: Azure Voice Live via
server/voiceLiveRelay.ts
- Auth: signed JWT session cookie; Google sign-in, password, or emailed one-time code (Resend)
Getting Started
1. Clone the repository
git clone https://github.com/your-username/liveinterviewtestapp.git
cd liveinterviewtestapp
2. Install dependencies
npm install
3. Configure Environment Variables
Copy the example environment file and fill in your API credentials:
cp .env.example .env.local
Key environment variables to configure in .env.local (see .env.example for the full list):
AUTH_SECRET (required): random string of at least 32 characters for signing sessions (openssl rand -base64 48)MONGODB_URI: MongoDB connection stringRESEND_API_KEY, EMAIL_FROM: send verification and password reset codes (in development, codes are printed to the server console when unset)ADMIN_EMAILS: comma-separated admin emails (admin rights apply once the address is verified)NEXT_PUBLIC_GOOGLE_CLIENT_ID: Google sign-inAZURE_SPEECH_KEY, AZURE_SPEECH_REGION: Nigerian TTS voices and realtime voice interviewsGEMINI_API_KEY: Google Gemini API keyELEVENLABS_API_KEY: ElevenLabs API keyYARNGPT_API_KEY: YarnGPT API key (for Nigerian localized accents)R2_ACCOUNT_ID, R2_ACCESS_KEY_ID, R2_SECRET_ACCESS_KEY, R2_PUBLIC_DOMAIN: Cloudflare R2 credentials
4. Run the development server
npm run dev
Open http://localhost:3000 with your browser.
For realtime voice interviews (/interview/realtime), also run the relay:
npm run voice-relay
Authentication
- Log in with a password, or with Continue with Google (the only passwordless option).
- Sign-up requires a password. It creates the account and emails a 6-digit verification code; no session is issued until the code is entered (
POST /api/auth/verify; resend with POST /api/auth/verify/resend).
- Forgot password:
POST /api/auth/password/forgot emails a reset code; POST /api/auth/password/reset with { email, code, password } sets the new password and signs the user in. Accounts without a password (e.g. Google sign-ups) can use this to add one.
- Codes expire after 10 minutes, allow 5 attempts, are single use, and are bound to their purpose (a verification code can't reset a password).
- Resetting or changing a password signs out every other session for that account (
sessionsValidAfter on the user).
- Login, sign-up, code sending and every paid API (LLM, TTS, realtime voice) are rate limited (
src/lib/rateLimit.ts).
- Older accounts may still hold plain-text passwords. Hash them once with
npx tsx scripts/migrate-plaintext-passwords.ts --apply (run without --apply for a dry run). Until then those users can use Forgot password.
Scripts
npm run dev: Start development servernpm run build: Build for productionnpm run start: Start production servernpm run lint: Run ESLint checksnpm test: Run unit tests (tests/)npm run typecheck: TypeScript checknpm run voice-relay: Start the realtime voice relaynpx tsx scripts/set-r2-cache-headers.ts [--apply]: Add the long-lived cache header to audio clips uploaded before uploads set it (one-off; dry run without --apply)npm run warm-tts-cache: Pre-warm TTS audio cache across questions, personas, and regions
