A web-based AI-powered platform that simulates real interview experiences using NLP, video analysis, and interactive dashboards. Designed to help students and job seekers prepare confidently for interviews while giving employers a data-driven screening tool.
👨💻 Built as a collaborative team project by 4 members. This repository showcases my personal involvement and contributions, particularly in machine learning model development and data gathering for AI training.
- Conduct mock interviews tailored to specific job roles and industries.
- Includes Easy, Medium, and Hard technical question categories.
- Uses Natural Language Processing (NLP) to analyze verbal responses.
- Provides real-time scoring and feedback based on tone, fluency, and relevance.
- Candidate responses recorded via webcam.
- Analyzed using CNNs (Convolutional Neural Networks) to detect emotions like happy, anxious, neutral, etc.
- Visual feedback helps users understand non-verbal cues.
- Built with ReactJS and Tailwind CSS.
- Candidates receive visualized analytics of their performance:
- Emotion timelines
- Response strength
- Confidence indicators
- Tracks user progress over multiple attempts.
- Dynamic question pool generated based on:
- Job description
- Domain knowledge
- Skill tags
-
✅ Led the effort to gather diverse, relevant interview questions and answers using
- ChatGPT, Reddit threads, and industry-specific resources
- Organized and labeled questions into categories Easy, Medium, Hard)
-
✅ Co-developed the emotion detection model using a CNN architecture
- Trained the model on labeled facial emotion datasets
- Integrated prediction pipeline into the feedback engine
-
✅ Assisted in validating model outputs against user responses during video-based interviews
-
✅ Collaborated with team to align ML outputs with frontend analytics and scoring
| Layer | Tech Used |
|---|---|
| Frontend | ReactJS, Tailwind CSS |
| Backend | Flask, PostgreSQL |
| AI/ML | TensorFlow, Scikit-learn, OpenCV |
| NLP & Audio | Speech-to-Text APIs, NLTK, TextBlob |
| Video Emotion | CNN Model (Kaggle-trained dataset) |
| Dashboards | Chart.js, React Hooks |
- 🔁 Gesture/posture analysis during video interviews
- 🌐 Multilingual feedback system
- 📲 Mobile-first interface for quick mock sessions
- 🧠 Gamified practice challenges and streaks
This project was developed as part of a 4-member team under the course Innovative Product Development.
Team members: Parth Das, Nilay Rathod, Neeharika Bhaide, Pooja Divekar
If you're interested in the project or want to collaborate on similar AI/ML tools, feel free to reach out: