IntervueAI

Live demo: https://salama-malek.github.io/IntervueAI/

Voice-first mock interview platform for practicing technical interviews in the browser.

Overview

IntervueAI runs a mock interview entirely through speech: it asks role- and seniority-specific questions, listens to your spoken answer via the Web Speech API, scores it heuristically (length, structure, and role-relevant keywords), and speaks back feedback before moving to the next question. It works fully client-side out of the box (no backend required), and can optionally be pointed at a REST API for session persistence and server-side scoring. Sessions are recorded as a live transcript, saved to history, and exportable as JSON.

Features

• Voice-first interaction using the Web Speech API for speech-to-text and text-to-speech
• Six interview tracks (Frontend, Backend, Full-Stack, Data Engineer, Product Manager, UX) across Junior/Mid/Senior levels
• Heuristic answer scoring based on response length, structure, and role keywords
• Live, diarized transcript with interim speech results
• Session history with detail view, JSON export, and delete
• Pluggable adapters: local heuristic engine or REST backend, switchable via settings or env vars
• Optional Mock Service Worker (MSW) layer to exercise the REST adapter without a live server
• Internationalization with automatic RTL layout (English, Arabic)
• Keyboard navigation, ARIA labels, and reduced-motion support

Tech stack

• React 18 + TypeScript, built with Vite
• Tailwind CSS
• Web Speech API (STT/TTS) and Web Audio API (microphone RMS detection)
• TanStack Query for data fetching
• Express backend (server/) exposing a minimal REST API, documented via server/openapi.yaml
• MSW for mocking the REST adapter in the browser
• Vitest + React Testing Library for tests; ESLint + Prettier for linting/formatting

Getting started

Prerequisites

• Node.js 18+
• Chrome or Edge (best Web Speech API support)
• HTTPS in production (required for microphone access)

Install

git clone <repository-url>
cd IntervueAI
npm install

Environment variables

Copy .env.example to .env.local and adjust as needed:

VITE_ADAPTER=local            # "local" (no backend) or "rest"
VITE_API_BASE_URL=http://localhost:8787
VITE_USE_MSW=false            # true to mock the REST adapter in-browser

Run (local adapter, no backend)

npm run dev

App runs at http://localhost:5173.

Run with the REST backend

npm run server:dev     # starts the Express API on port 8787
npm run dev:full       # runs the API and Vite dev server together

Set VITE_ADAPTER=rest and VITE_API_BASE_URL=http://localhost:8787 in .env.local, or switch adapters from the Adapter dropdown in the app's settings panel.

Scripts

npm run typecheck      # TypeScript type checking
npm run lint           # ESLint
npm run format         # Prettier
npm run test           # Vitest
npm run test:coverage  # Vitest with coverage
npm run build          # Production build
npm run preview        # Preview the production build

Project structure

IntervueAI/
├── src/
│   ├── adapters/         # Local (heuristic) and REST interview adapters
│   ├── components/       # UI: AppShell, InterviewRoom, TranscriptList, HistoryPanel, ReportPanel, ...
│   ├── context/          # Auth context/state
│   ├── hooks/            # useSpeechStt, useSpeechTts, useRmsGate, useAuth, useSessionsData
│   ├── lib/               # Question bank, scoring, summarization, i18n, API client
│   ├── mocks/             # MSW handlers for the REST adapter
│   └── App.tsx, main.tsx
├── server/                # Express REST API + OpenAPI spec
├── examples/              # Example exported sessions
└── package.json

License

MIT