InterviewSimAI

Train smarter. Interview sharper.

A React Native + Expo mobile app that runs realistic AI-driven mock interviews end to end. Pick a track, get questions generated by Gemini, record spoken answers (transcribed by Gemini), and finish with a scored AI feedback report — all on device, with sessions persisted locally.

Screenshots

Home	Live Interview	Recording & Transcript

Score & Subscores	Strengths & Improvements

Features

4 interview tracks — React Native, HR / Behavioral, DSA, System Design.
AI-generated questions — Gemini produces fresh, schema-validated questions per session, with a built-in fallback to a curated sample set if the API is unreachable.
Voice answers + AI transcription — Tap to record, Gemini transcribes the audio, and you can still edit the text before submitting.
Live session timer with pause / resume.
Holistic AI feedback — overall score (0–100) plus subscores for technical knowledge, communication, and confidence, a written summary, 3–5 strengths and 3–5 concrete improvements.
Local session history — every completed interview is saved to AsyncStorage so you can revisit past feedback.
Polished dark UI with NativeWind, gradient backgrounds, animated transitions, and a score ring that adapts its color to performance.
Cancellable AI requests — leaving a screen or retrying aborts in-flight Gemini calls cleanly.
Resilient networking — automatic retries with exponential backoff on 429 / 5xx responses.

Tech Stack

Layer	Choice
Runtime	Expo SDK 56, React Native 0.85, React 19
Language	TypeScript (strict mode, `@/*` path alias)
Styling	NativeWind v4 (Tailwind CSS for RN)
Navigation	`@react-navigation/native-stack` v7
State	Zustand + `persist` middleware
Storage	`@react-native-async-storage/async-storage`
Audio	`expo-audio` (new SDK 56 API)
File I/O	`expo-file-system` (new `File` API)
AI	Google Gemini (`gemini-2.5-flash` by default) via `axios`
Animations	`react-native-reanimated` v4
Icons	`@expo/vector-icons` (Ionicons)

Getting Started

Prerequisites

Node.js 20+
npm (or your preferred package manager)
Expo Go on a physical device, or Xcode (iOS Simulator) / Android Studio (Emulator)
A Google Gemini API key from Google AI Studio

Note: Audio recording does not work on the iOS Simulator. Use a real device or an Android emulator with a working mic for the recording flow.

1. Install

git clone <repo-url>
cd InterviewSimAI
npm install

2. Configure environment

Create a .env file in the project root (use .env.example as a template):

EXPO_PUBLIC_GEMINI_API_KEY=your-gemini-api-key
EXPO_PUBLIC_GEMINI_MODEL=gemini-2.5-flash

EXPO_PUBLIC_* variables are inlined into the JS bundle at build time, so restart Expo after changing .env.

3. Run

npm start          # Expo dev server (pick a target from the menu)
npm run ios        # iOS Simulator
npm run android    # Android Emulator
npm run web        # Web build

How It Works

Home ──▶ Interview ──▶ Feedback
 │           │             │
 │           ├─ Gemini: generateInterviewQuestions
 │           ├─ expo-audio: record answer
 │           ├─ Gemini: transcribeAudioRecording
 │           └─ Gemini: evaluateSession (on finish)
 │
 └─ Recent session from local history

Home — choose a category and tap Start.
Interview — on mount, the app requests 3 fresh questions from Gemini. The "AI interviewer" speaks the question, then you tap the mic, record your answer, and Gemini transcribes it. You can edit the transcript before submitting.
Feedback — after the last question, the full Q&A is sent to Gemini for a holistic evaluation. The session is saved to local history and you land on the feedback screen.

Project Structure

.
├── App.tsx                    # Root: providers + navigator + history hydration
├── index.ts                   # registerRootComponent(App)
├── app.json                   # Expo config (incl. expo-audio mic permission)
├── babel.config.js            # babel-preset-expo + nativewind/babel
├── metro.config.js            # withNativeWind(...)
├── tailwind.config.js         # Design tokens (brand / accent / ink palettes)
├── global.css                 # Tailwind directives
├── .env.example               # Required env vars
├── assets/                    # App icons, splash, favicon
├── docs/screenshots/          # README screenshots
└── src/
    ├── components/            # 13 reusable UI components (cards, buttons, waveform, …)
    ├── constants/             # theme, layout helpers, mockQuestions fallback
    ├── hooks/
    │   └── useAudioRecorder.ts  # expo-audio wrapper (permissions, mode, lifecycle)
    ├── navigation/
    │   └── AppNavigator.tsx     # Native stack: Home → Interview → Feedback
    ├── screens/
    │   ├── HomeScreen.tsx
    │   ├── InterviewScreen.tsx
    │   └── FeedbackScreen.tsx
    ├── services/
    │   ├── aiService.ts         # Gemini client (questions, transcription, feedback)
    │   ├── historyService.ts    # Persisted session history
    │   └── storage.ts           # Typed AsyncStorage wrapper
    ├── store/
    │   └── interviewStore.ts    # Zustand store + selectors
    ├── types/                   # Domain + navigation types
    └── utils/
        └── score.ts             # Score tone → color/label maps

Gemini Integration (`src/services/aiService.ts`)

The AI service is a single, dependency-light module that:

Reads EXPO_PUBLIC_GEMINI_API_KEY / EXPO_PUBLIC_GEMINI_MODEL and throws a descriptive GeminiServiceError if the key is missing.
Uses Gemini's JSON response schemas (responseMimeType: 'application/json' + responseSchema) so questions and feedback come back as strict, typed JSON — no fragile regex parsing.
Sends audio inline via inlineData (base64), capped at 14 MB to stay within the inline-data limit.
Retries on 429 / 500 / 502 / 503 / 504 with exponential backoff (up to 3 retries).
Supports AbortSignal for cancel-on-unmount and retry semantics.

Exposed helpers:

Function	Purpose
`generateInterviewQuestions`	Topic + difficulty → array of questions
`transcribeAudioRecording`	Audio file URI → plain-text transcript
`evaluateAnswer`	Single Q/A → scored feedback
`evaluateSession`	Full Q/A list → overall + subscored feedback
`isGeminiConfigured`	Quick env-var check

State Management

A single Zustand store (src/store/interviewStore.ts) owns:

Selection: selectedCategoryId, selectedDifficulty
Live session: currentSessionId, currentQuestions, currentQuestionIndex, currentAnswers, transcript
History: history, historyHydrated (mirrored to historyService)
Loading flags: questions | transcription | feedback
Timer: { isRunning, elapsedMs, startedAt } with start / pause / tick / reset

Only the small selection slice is persisted by the store middleware (partialize). The heavier history blob is owned by historyService, which serializes through a typed, namespaced AsyncStorage wrapper (@interviewsim:interview-history/v1).

Exported selectors: selectCurrentQuestion, selectProgress, selectElapsedMs, selectAverageScore, plus a formatElapsed(ms) helper.

Available Scripts

npm start        # expo start
npm run ios      # expo start --ios
npm run android  # expo start --android
npm run web      # expo start --web

Troubleshooting

"Missing EXPO_PUBLIC_GEMINI_API_KEY" — Add the key to .env and restart expo start (Metro caches env vars).
No questions appear / "Couldn't load AI questions" — Tap Use sample set in the error banner to fall back to bundled questions, or check your API key / quota.
"Recording is not supported on the iOS Simulator" — Use a real device or an Android emulator.
"Recording is too large to transcribe inline" — Keep individual answers under ~14 MB of audio; the app prompts you to record a shorter answer.
History feels stale — The store hydrates history once on app start (App.tsx). Fully reload the app or clear history from within the store to reset.

Security Notes

EXPO_PUBLIC_* env vars are bundled into the client app. This is fine for local development and personal use, but if you ship this anywhere public you should:

Proxy Gemini calls through a backend you control.
Rotate the key in Google AI Studio if it ever leaks.
Restrict the key (HTTP referrers / API restrictions) wherever possible.

License

This project is provided as-is for learning and personal use. Add a license of your choice before distributing.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

InterviewSimAI

Screenshots

Features

Tech Stack

Getting Started

Prerequisites

1. Install

2. Configure environment

3. Run

How It Works

Project Structure

Gemini Integration (`src/services/aiService.ts`)

State Management

Available Scripts

Troubleshooting

Security Notes

License

About

Uh oh!

Releases

Packages

Uh oh!

Contributors

Uh oh!

Languages

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
assets		assets
docs/screenshots		docs/screenshots
src		src
.env.example		.env.example
.gitignore		.gitignore
App.tsx		App.tsx
LICENSE		LICENSE
README.md		README.md
app.json		app.json
babel.config.js		babel.config.js
global.css		global.css
index.ts		index.ts
metro.config.js		metro.config.js
nativewind-env.d.ts		nativewind-env.d.ts
package-lock.json		package-lock.json
package.json		package.json
tailwind.config.js		tailwind.config.js
tsconfig.json		tsconfig.json

Folders and files

Latest commit

History

Repository files navigation

InterviewSimAI

Screenshots

Features

Tech Stack

Getting Started

Prerequisites

1. Install

2. Configure environment

3. Run

How It Works

Project Structure

Gemini Integration (src/services/aiService.ts)

State Management

Available Scripts

Troubleshooting

Security Notes

License

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Gemini Integration (`src/services/aiService.ts`)

Packages