🎤 Composable AI Mock Interviewer

🚀 Live Demo

👉 https://composable-ai-mock-interviewer-pi.vercel.app/

An AI-powered mock interview platform that simulates real interviews (HR / Tech / Behavioral) with adaptive questioning, live evaluation, audio mode, and a final performance report — built for rapid MVP demos and hackathons.

This project enables a candidate to:

1. Choose interview type & role
2. Provide skills or resume context
3. Answer adaptive questions
4. Get instant scores + coaching tips
5. Receive a final structured interview report

Designed with a Composable Blocks mindset where each stage (Question, Memory, Evaluation, Feedback, Report) is modular and reusable.

---

🚀 Features

🧩 Composable Pipeline

• Role Block
• Question Generator
• Memory Block
• Evaluation Block
• Feedback Block
• Final Report Block

Each block is independent and reusable — perfect for rapid prototyping and AI workflows.

🧠 Adaptive Interview Logic

Each next question adapts based on:

• Previous answer quality
• Weak / strong skills
• Role & interview type

This makes every interview feel real and personalized.

📊 Live Evaluation

• Scores for: Clarity, Relevance, Depth, Confidence
• Overall score per answer
• Strengths & improvement areas in real time

🎧 Audio Mode

• 🔊 Speak questions (Text-to-Speech)
• 🎙️ Answer using mic (Speech-to-Text)

📄 Final Interview Report

• Overall score
• Skill-wise breakdown
• Top strengths & gaps
• 7-day improvement plan

---

🛠 Tech Stack

• Next.js (App Router) – Frontend + Backend in one app
• LLM – Gemini / OpenAI (pluggable adapter)
• In-memory Store – Fast MVP sessions
• Web Speech API – Audio input/output

---

📁 Project Structure

ai-mock-interviewer/
├── app/
│   ├── page.tsx              # Setup screen
│   ├── interview/
│   │   └── page.tsx          # Interview UI
│   └── report/
│       └── page.tsx          # Final report
│
├── api/
│   └── session/
│       ├── create/
│       │   └── route.ts
│       ├── next-question/
│       │   └── route.ts
│       ├── submit-answer/
│       │   └── route.ts
│       └── report/
│           └── route.ts
│
├── lib/
│   ├── store.ts
│   ├── llm.ts
│   ├── prompts.ts
│   ├── schema.ts
│   └── scoring.ts
│
├── components/
│   ├── BlocksPanel.tsx
│   ├── ChatPanel.tsx
│   ├── OutputPanel.tsx
│   └── AudioControls.tsx
│
├── README.md
├── next.config.js
└── package.json

---

⚙️ Setup

1. Create project

npm create next-app@latest ai-mock-interviewer --ts
cd ai-mock-interviewer
cd ai-mock-interviewer

Add environment key

Create .env.local:

GEMINI_API_KEY=your_api_key_here
GEMINI_MODEL=gemini-2.0-flash

Run

npm run dev
Open: http://localhost:3000

🧪 Demo Flow (For Judges) Select Tech Interview → Data Analyst

Enter skills: SQL, Power BI, Python

Start interview

Answer Q1 via mic

Show:

Live scores

Tips

Memory update

Next question adapts based on weakness

Complete 5 turns

Show Final Report

---

🎯 Why This Stands Out Clear end-to-end user journey

Shows real AI reasoning + memory

Demonstrates composability

Works in 24-hour hackathon scope

Easy to extend into a real product

---

🗺 Roadmap

🌗 Dark / Light Mode Toggle

📄 Resume Upload & Resume-Based Questions

📊 User Dashboard & Progress Tracking

📥 Downloadable PDF Report

---

👨‍💻 Author

Built by Jitendra Gaherwar

B.Tech IT | Data & AI Developer

Focused on building practical AI tools for learning and productivity.