🎯 InterviewAI — AI Mock Interview Coach

A full-stack Next.js app for AI-powered mock interviews. 100% free, no subscription required.

Uses:

• HuggingFace Router API (router.huggingface.co) for all AI inference
• Mistral-7B-Instruct via Nebius provider for question generation and feedback
• OpenAI Whisper Large v3 via HF Inference provider for speech-to-text
• Web Speech API as real-time live-transcript fallback
• WebRTC MediaDevices for webcam preview

⚠️ Note: The old api-inference.huggingface.co endpoint is deprecated. This app uses the new router.huggingface.co endpoint with the OpenAI-compatible chat completions format.

---

🚀 Quick Start

1. Install dependencies

npm install

2. Set up your HuggingFace token

Get a free token at https://huggingface.co/settings/tokens (free account, no credit card needed)

cp .env.local.example .env.local
# Edit .env.local:
# HUGGINGFACE_API_TOKEN=hf_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Users can also paste their token directly in the app UI — it's never stored, only kept in sessionStorage for the active session.

3. Run

npm run dev

Open http://localhost:3000

---

🔌 API Endpoints Used

Route	HF Endpoint	Model
Text generation	router.huggingface.co/nebius/v1/chat/completions	Mistral-7B-Instruct-v0.3
Fallback generation	router.huggingface.co/novita/v1/chat/completions	Mistral-7B-Instruct-v0.3
Speech-to-text	router.huggingface.co/hf-inference/models/openai/whisper-large-v3	Whisper Large v3

---

📁 Project Structure

interview-ai/
├── app/
│   ├── page.tsx                          # Setup screen
│   ├── interview/page.tsx                # Live interview
│   ├── results/page.tsx                  # Final report
│   └── api/
│       ├── generate-questions/route.ts   # Mistral question generation
│       ├── evaluate-answer/route.ts      # Per-answer feedback
│       ├── transcribe/route.ts           # Whisper STT
│       ├── final-report/route.ts         # Final assessment
│       └── parse-cv/route.ts             # PDF parsing
├── components/
│   ├── WebcamFeed.tsx
│   └── AudioRecorder.tsx
└── lib/
    ├── hf.ts        # HF Router helpers (no SDK, plain fetch)
    └── store.ts     # Session state

---

🎤 Features

Feature	Tech
CV Upload & Parsing	pdf-parse (server-side)
AI Question Generation	Mistral-7B via HF Router
Speech-to-Text	Whisper Large v3 via HF Router
Live Transcript	Web Speech API
Webcam	WebRTC MediaDevices
Per-answer Feedback	Mistral-7B scoring & coaching
Final Report	Mistral-7B comprehensive assessment
Text-to-Speech	Web Speech Synthesis API
Audio Meter	Web Audio API AnalyserNode

---

📦 Production Build

npm run build && npm start