PrepAI – Multimodal AI Mock Interview Platform

🚀 Overview

PrepAI is an AI-powered mock interview platform that simulates real-world interview environments using Generative AI, Speech Processing, NLP, and Computer Vision.

Unlike traditional chatbot-based tools, PrepAI evaluates not only technical responses but also communication quality, confidence, posture, gaze stability, tone clarity, and nervousness — delivering a comprehensive performance report.

---

🎯 Problem Statement

Interview preparation often suffers from:

• Limited access to expert mock interviewers
• Subjective and inconsistent feedback
• No evaluation of body language or speech patterns
• Static, non-adaptive questioning systems

PrepAI addresses these gaps using a multimodal AI-driven evaluation framework.

---

🧠 Core Features

1️⃣ Resume-Based Personalization

• Extracts resume text using PyPDF2 and SpaCy
• LLM analyzes strengths and skill gaps
• Generates role-specific and difficulty-adjusted questions

2️⃣ Real-Time Adaptive Interview Engine

• Role-based interviews (ML Engineer, Data Analyst, HR, etc.)
• Whisper-based Speech-to-Text transcription
• LLM-driven answer evaluation
• Dynamic follow-up question generation
• gTTS-based spoken question simulation

3️⃣ Audio Intelligence

• Speech rate and fluency scoring
• Tone stability (jitter detection)
• Nervousness estimation via voice stress patterns

4️⃣ Computer Vision Analysis

• MediaPipe-based posture tracking
• Eye-gaze tracking and blink-rate monitoring
• CNN-based emotion recognition

5️⃣ Multimodal Scoring Engine

Weighted evaluation rubric:

• Technical Accuracy — 45%
• Communication Quality — 20%
• Body Language — 15%
• Attention & Engagement — 10%
• Nervousness Score — Heuristic (Audio + CV signals)

Generates:

• Full interview transcript
• Behavioral metrics
• Structured performance report
• Personalized improvement roadmap

---

🏗 Technical Architecture

Frontend → FastAPI Backend → AI Processing Modules

Backend

• FastAPI
• MongoDB

AI Components

• LLM (OpenAI / Gemini) for evaluation & adaptive questioning
• Whisper STT for transcription
• MediaPipe + OpenCV for posture & gaze tracking
• Custom CNN for emotion detection
• Audio feature extraction pipeline

Deployment

• Vercel / Render
• Git for version control

---

⚙️ Challenges & Mitigations

Posture Detection Accuracy

• Smoothing filters
• Diverse training data
• Validation layers

Real-Time Coding Evaluation

• Rule-based validation
• Test-case verification
• Content filtering

Scalability & Performance

• Cloud deployment
• Secure data handling
• Optimized API response time

---

🔮 Future Enhancements

• Cheat detection (plagiarism + gaze diversion detection)
• Real-time interrupting AI for off-track responses
• Company-specific interview simulations
• Advanced analytics dashboard

---

📌 Impact

PrepAI bridges the gap between technical evaluation and behavioral analysis by combining NLP, speech processing, and computer vision into a unified AI interview simulator.

It delivers objective, scalable, and personalized interview preparation at scale.