🤖TakeInt-AI Mock Interview Platform

Practice technical & behavioral interviews with AI voice agents, instant feedback, and performance analytics.

Tech Stack: Next.js • React • TypeScript • Prisma • NextAuth • Google Gemini API • Vapi (voice agents) • shadcn/ui • Tailwind CSS

---

✨ Overview

AI Mock Interview Platform helps candidates rehearse interviews with realistic voice-driven agents and structured question sets. It generates tailored questions (role, level, skills), records responses (voice), scores performance, summarizes strengths & gaps, and tracks progress over time.

• Voice Interviews: Real-time AI voice via Vapi (web calling + event webhooks).
• Question Generation: Google Gemini for role-aware question sets.
• Feedback & Summaries: Gemini-based rubric, strengths, and next steps.
• Authentication: NextAuth with Google Sign-In + Credentials (demo accounts supported).
• UI: shadcn/ui components with Tailwind CSS.
• Data: Prisma ORM (PostgreSQL by default).
• Pages: Landing, Interview Studio, History, Dashboard (Skill & Session analytics), Settings.

---

📸 Media

Landing Page	Dashboard — Skills Performance
Mock Interview Studio	Interview Panel

🎥 Video Walkthrough
▶️ Watch the demo video here

---

🧱 Architecture

• App Router (Next.js) with server actions for secure data ops.
• NextAuth for sessions (JWT) with Google provider + Credentials provider.
• Prisma schema with User, Interview, Question, Answer, Score,Transcript.
• Vapi session init on the client, server-side webhook (/api/vapi/webhook) for call events & transcripts.
• Gemini used server-side for question generation, summaries, and feedback .
• shadcn/ui for accessible, consistent components; charts via a lightweight React chart lib.

---

✅ Features

• Role-based Templates: Option to create custom mock interviews.
• Interview Modes: Behavioral, Algo/DSA, System Design, Mixed.
• Voice : Speak with the agent .
• Scoring: Evaluated across key dimensions — Problem Solving, System Design, Communication Skills, Technical Accuracy, Behavioral Responses, and Time Management.
• Feedback: Bullet tips, improvement plan, follow-up questions.
• History: All sessions with transcripts & audio replays.
• Dashboard: Skill radar, progress over time, topic heat-map.

---

🔧 Requirements

• Node.js 18+ (suggested 20+)
• PNPM / Yarn / NPM (examples use pnpm)
• Database: PostgreSQL (default)
• Environment variables configured (see below)

---

🔐 Environment Variables

Create .env.local in the project root:

# NextAuth
AUTH_SECRET=
AUTH_GOOGLE_ID=
AUTH_GOOGLE_SECRET=

# URLs
NEXT_PUBLIC_BASE_URL=http://localhost:3000

# Voice Agents (Vapi)
NEXT_PUBLIC_VAPI_TOKEN=

# Database (Postgres example)
DATABASE_URL=

# Google Gemini
GOOGLE_GENERATIVE_AI_API_KEY=

Tip: Generate AUTH_SECRET with openssl rand -base64 32

---

🗣️ Voice (Vapi) Integration

• Client: Initialize Vapi with NEXT_PUBLIC_VAPI_TOKEN; start/stop calls from the Interview .
• Server: POST /api/vapi/webhook receives call events, partial/final transcripts, and stores them via Prisma.
• Sync: On-call end, the transcript triggers Gemini summarization + scoring.

---

🧠 Google Gemini Usage

• Question Generation: Creates interview questions based on role, level, and skills.
• Feedback: Provides detailed interview feedback — scores, strengths, improvements, and follow-up suggestions.

---

🧭 Key Pages

• Landing – Marketing info, feature highlights, and call-to-action (sign in / get started).
• Dashboard – Skill radar, recent improvements, session trends, and quick actions.
• Create Interview – Build a custom mock interview (role, skills, level, mode).
• Mock Interview List – View all created mock interviews and select one to take.
• Interview Panel – Live interview space (voice and real-time Q&A).
• History – Past sessions with transcripts and feedback.

---

Made with ❤️ using Next.js, TypeScript, Prisma, NextAuth, shadcn/ui, Vapi, and Google Gemini.