An AI-powered interview assistant to help you ace your next opportunity — built in 36 hours at WildHacks 2025 by Northwestern University.
InView is a smart, real-time interview assistant that simulates technical interviews, evaluates your answers, and provides emotion-based feedback. It's designed to help job seekers build confidence, improve communication, and get actionable insights into their performance.
-
AI-Generated Interview Questions
Automatically tailored to the job description or role (e.g., React Developer, Data Scientist). -
Real-time Answer Evaluation (Gemini API)
Evaluates responses with detailed scoring, feedback, and suggestions. -
Live Emotion Detection via Webcam (FER)
- Analyzes facial expressions frame-by-frame.
- Tracks stress level, confusion, confidence, focus, and presence.
-
Post-Interview Summary Dashboard
- Emotion metrics (stress/confidence/confusion/focus).
- AI-based technical evaluation.
- Replay session and review answers.
-
Custom JD Input + Role Selector Choose predefined job roles or paste your own job description.
-
Privacy-First Approach
All video processing is local unless the user chooses to upload.
| Layer | Technology |
|---|---|
| Frontend | React.js, MUI (Material UI) |
| Backend | FastAPI (Python) |
| AI/ML APIs | Google Gemini API, FER Library |
| Video/Emotion | WebRTC, MediaRecorder API, OpenCV |
| Database (Planned) | MongoDb Atlas |
- User selects a job role or enters a custom JD
- LLM (Gemini) generates tailored technical questions
- User answers in text while webcam records live video
- Emotion snapshots captured every second via FER
- After interview ends:
- Answers are evaluated by Gemini
- Emotion data analyzed
- Final feedback is displayed with insights and scores
- Mohammed Saalim Kartapillai
- Derick Johnson
- Eric Somogyi
- Salvador Ortiz
# Frontend
cd interview-ai-frontend
npm install
npm start
# Backend
cd backend
pip install -r requirements.txt
uvicorn main:app --reload
Backend runs on http://localhost:8000
Frontend runs on http://localhost:3000