AI Interview Assistant — Human-in-Loop System with Complete State Management and Rollback

🧠 Problem Alignment

This project addresses the “Human-in-Loop System with Complete State Management and Rollback” problem under the Fullstack Engineer track.

Modern AI systems often require real-time human feedback before continuing a workflow.
The AI Interview Assistant implements this concept by placing a human (the candidate) inside an interactive, event-driven interview system powered by AI.

---

🎯 Overview

The AI Interview Assistant is an end-to-end web application that:

• Parses the candidate’s resume
• Dynamically generates interview questions using the Google Gemini API
• Simulates a live interview with timed responses
• Evaluates answers asynchronously through AI
• Provides immediate feedback, scoring, and performance analytics

The system is event-driven, stateful, and resilient — capable of pausing/resuming workflows, handling delays, and supporting retries or feedback loops.

---

⚙️ Architecture Overview

🧩 Core Components

1. Frontend (React + Vite + Ant Design)

   • Dynamic, configurable UI for interview flow
   • Tracks current question, timer, and user responses
   • Communicates asynchronously with backend APIs

2. Backend (Node.js + Express)

   • Handles orchestration logic and API routes
   • Integrates with Google Gemini API for AI-driven Q&A and feedback
   • Maintains interview state and logs for replay or retry

3. State Management & Rollback

   • Interview session stored in memory or database
   • Supports reattempting questions and restarting sessions
   • Logs AI responses and timestamps for observability

4. Observability & Feedback

   • Dashboard visualizes interview performance and AI feedback
   • Each session includes analytics: time spent, question category, score, and AI evaluation

---

🛠️ Tech Stack

Layer	Technology
Frontend	React, Vite, Ant Design
Backend	Node.js, Express
AI Model	Google Gemini API
State Management	Session store + local state tracking
Deployment	Render (Monorepo setup)

---

💡 Key Features

• Real-time human–AI interaction (Human-in-Loop)
• Stateful session management and dynamic resume/rollback
• Asynchronous AI evaluation and feedback
• Event-driven API design (frontend & backend decoupled)
• Fully deployable on Render or any Node/React host

---

⚙️ Evaluation Mapping

Evaluation Criteria	Implementation
Architecture Thinking (25%)	Event-driven design; separate AI orchestration backend
State Model Design (20%)	Interview progress tracked with states: pending → evaluating → completed
Frontend Configurability (15%)	Dynamic React components render questions, timers, and feedback
Resilience & Retry Logic (15%)	Supports session restarts and reattempts
Integration Creativity (15%)	Gemini API integration for multi-round interview logic
Observability & Feedback Loop (10%)	Performance dashboard and detailed answer analytics

---

🧱 Repository Structure

ai-interview-assistant/
│
├── backend/ # Express backend
│ ├── server.js # Main backend entry point
│ ├── package.json # Backend dependencies
│ ├── .env # Backend environment variables (not committed)
│ └── public/ # Built frontend assets (copied from frontend/dist)
│
├── frontend/ # React + Vite frontend
│ ├── src/ # Frontend source code
│ │ ├── api.js # API wrapper for backend calls
│ │ ├── App.jsx # Main React App
│ │ ├── pages/ # Pages (InterviewPage, DashboardPage, etc.)
│ │ └── components/ # Reusable React components
│ ├── package.json # Frontend dependencies
│ ├── .env # Frontend environment variables
│ └── dist/ # Production build (generated by Vite)
│
├── .gitignore # Ignore node_modules, env, build files
├── README.md # Project documentation
└── render.yaml # (Optional) Render configuration

---

🚀 Demo

Live App: [https://your-ai-interview-assistant-demo-link.com](#)
GitHub: https://github.com/Guna1610/ai-interview-assistant

---

🔮 Future Enhancements

• Add multi-agent evaluation (HR bot + Tech bot)
• Integrate Slack/Gmail notifications for interview summaries
• Enable long-term analytics with MongoDB or Firebase
• Add Lyzr SDK agent orchestration integration

---

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/Guna1610/ai-interview-assistant.git
cd ai-interview-assistant

2. Setup Backend

cd backend
npm install

• Create a .env file in backend/ with:
• GEMINI_API_KEY=your_google_gemini_api_key
• NODE_ENV=production

Run locally:

node server.js

3. Setup Frontend

cd ../frontend
npm install

• Create .env in frontend/:
• VITE_API_URL=http://localhost:4000

Run locally:

npm run dev

• Visit: http://localhost:5173

🌐 Deployment on Render

This repo is already configured for monorepo deployment:

Build Command:

cd frontend && npm ci && npm run build && cd ../backend && npm ci && mkdir -p public && cp -r ../frontend/dist/* public/

Start Command:

node server.js

Environment Variables (Render Dashboard → Environment):

• GEMINI_API_KEY=your_google_gemini_api_key
• NODE_ENV=production
• VITE_API_URL=https://your-service-name.onrender.com
• Render automatically serves the React build from /backend/public and proxies API calls to Express.

---

📸 Screenshots

---

🤝 Contributing

• Fork the repo
• Create a new branch (feature-xyz)
• Commit changes and push
• Open a Pull Request

---

🚀 Demo

Live App: https://your-ai-interview-assistant-demo-link.com
GitHub: https://github.com/Guna1610/ai-interview-assistant

---

🔮 Future Enhancements

• Add multi-agent evaluation (HR bot + Tech bot)
• Integrate Slack/Gmail notifications for interview summaries
• Enable long-term analytics with MongoDB or Firebase
• Add Lyzr SDK agent orchestration integration

---

📜 License

MIT License — free to use and modify