Created & Developed by Gautam Kumar | Full-Stack Developer & AI Engineer
Production-grade AI interview simulation platform with real-time ML proctoring, CRDT collaborative coding, multimodal candidate analysis, and WebRTC video — serving enterprise-level candidate evaluation at scale.
InterviewMinds is designed, architected, and built by Gautam Kumar — a Full-Stack Developer and AI Engineer specializing in enterprise-grade web applications with machine learning integration.
- 🌐 Portfolio: gautam-kr.vercel.app
- 💼 LinkedIn: linkedin.com/in/gautamkr62
- 🐙 GitHub: github.com/theunstopabble
- 📧 Email: gautamkumar43421@gmail.com
┌─────────────────────────────────────────────────────────────────────────────┐
│ INTERVIEWMINDS │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌───────────────────────────────────────────────────────────────────────┐ │
│ │ FRONTEND — React 18 + Vite 7.3 + Tailwind CSS + Shadcn UI │ │
│ │ ├── Clerk Auth (OAuth/Email) ├── Monaco Code Editor │ │
│ │ ├── face-api.js (Browser ML) ├── Socket.IO Client │ │
│ │ ├── Recharts (Analytics) └── PWA Support │ │
│ └───────────────────────────────────────────────────────────────────────┘ │
│ │ HTTPS + JWT + WebSocket │
│ ┌───────────────────────────────────────────────────────────────────────┐ │
│ │ BACKEND — Node.js 20 + Express 4.21 + TypeScript 5.3 │ │
│ │ ├── 81 Service Modules ├── 38 REST Route Files │ │
│ │ ├── GraphQL (Apollo Server 4) ├── Socket.IO 4.8 (Real-time) │ │
│ │ ├── BullMQ Workers ├── Circuit Breaker Pattern │ │
│ │ └── ML Pipeline (face-api.js + TensorFlow.js) │ │
│ └───────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────────────────────────┐ │
│ │ MongoDB │ │ Redis │ │ External Services │ │
│ │ Atlas │ │ (Upstash) │ │ ├── Groq (Llama 3.3-70b LLM) │ │
│ │ │ │ │ │ ├── Azure Speech (TTS/STT) │ │
│ │ 7 Models │ │ Cache │ │ ├── Piston (Code Execution) │ │
│ │ Pool: 20 │ │ BullMQ │ │ ├── Cloudinary (Media) │ │
│ │ │ │ Rate Limit │ │ └── WebRTC (STUN/TURN) │ │
│ └──────────────┘ └──────────────┘ └──────────────────────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────────────┘
InterviewMinds implements 91 enterprise features across 18 development phases, all production-ready with real API integrations (no mock data):
| Phase | Domain | Key Features |
|---|---|---|
| 1 | Core Interview Engine | AI-powered chat (Groq Llama 3.3-70b), resume parsing (pdf2json + vector embeddings), dynamic follow-up questions |
| 2 | Enterprise Security | E2E encryption (RSA-4096 + AES-256-GCM + PBKDF2), biometric auth, fraud detection, geo-fencing |
| 3 | Video Proctoring | ML face detection (face-api.js), eye tracking via landmarks, expression analysis, screen monitoring |
| 4 | Smart Assessment | Sandboxed code execution (Piston API), answer validation, job matching, skill gap analysis |
| 5 | Integrations | SSO (OAuth/SAML), webhooks (HMAC-signed), ATS integration (Greenhouse/Lever) |
| 6 | Enterprise Ops | Multi-tenancy, RBAC (candidate/interviewer/admin), tenant isolation |
| 7 | Analytics | MongoDB aggregation pipelines, trend analysis, top performers, pipeline tracking |
| 8 | Scheduling & Reports | Calendar booking, timezone handling, question bank, scorecards, PDF/CSV exports |
| 9 | Next-Gen AI | LLM interviewer with memory/personas, multimodal voice tone analysis (Groq primary), smart adaptive assessment |
| 10 | Infrastructure | Circuit breakers, connection pooling, CDN integration, graceful shutdown |
| 11 | Real-Time Collaboration | CRDT collaborative editor (three-way merge), whiteboard (MongoDB persistence), video calls (WebRTC + STUN/TURN + SFU) |
| 12 | Predictive Analytics | Attrition prediction, performance forecasting, anomaly detection, sentiment analysis |
| 13 | Developer Experience | GitHub/GitLab integration, code review, sandboxed execution, test runner |
| 14 | Compliance & Governance | GDPR data export, PII masking, data retention policies, immutable audit trails |
| 15 | Integration Ecosystem | HRIS (Workday/BambooHR/SAP), Slack/Teams/Discord, Zoom/Google Meet, background checks |
| 16 | Mobile & Multi-Platform | Offline support, push notifications, responsive mobile API |
| 17 | Observability | Prometheus metrics, Pino structured logging, Sentry error tracking, health checks |
| 18 | AI Agents & Automation | Autonomous screening/scheduling/feedback agents, workflow automation engine |
| Layer | Technology | Version |
|---|---|---|
| Runtime | Node.js | 20.x |
| Language | TypeScript | 5.3 |
| Frontend | React + Vite + Tailwind CSS + Shadcn UI | 18 / 7.3 / 3.4 |
| Backend | Express + Socket.IO + Apollo Server (GraphQL) | 4.21 / 4.8 / 4.11 |
| Database | MongoDB (Mongoose ODM) | 8.x |
| Cache & Queue | Redis (ioredis) + BullMQ | 5.10 / 5.45 |
| Monorepo | Turborepo + npm workspaces | latest |
| Service | Model/Technology | Purpose |
|---|---|---|
| Groq | Llama 3.3-70b-versatile | Interview chat, voice analysis, resume screening, AI agents |
| face-api.js | SSD MobileNet + Landmarks + Expressions | Real-time face detection & proctoring |
| TensorFlow.js | Runtime engine | Server-side ML model execution |
| Azure Speech SDK | Neural TTS/STT | Text-to-speech for AI interviewer voice |
| Piston API | Multi-language sandbox | Secure code execution (20+ languages) |
| Xenova Transformers | all-MiniLM-L6-v2 | Resume text embeddings for semantic search |
| Feature | Implementation |
|---|---|
| Authentication | Clerk (OAuth + Email + Magic Links) |
| Authorization | Custom RBAC with wildcard admin permissions |
| Data Encryption | RSA-4096 + AES-256-GCM + PBKDF2 key derivation |
| API Security | Helmet CSP + Rate Limiting + CSRF + Input Sanitization |
| Audit Trail | Immutable logs with correlation IDs and IP tracking |
InterviewMinds/
├── apps/
│ ├── api/ # Backend API Server
│ │ ├── src/
│ │ │ ├── config/ # Service configurations
│ │ │ ├── graphql/ # Apollo Server schema + resolvers
│ │ │ ├── lib/ # 81 business logic service modules
│ │ │ │ ├── aiAgent.ts # AI agent orchestration (Groq)
│ │ │ │ ├── analytics.ts # MongoDB aggregation analytics
│ │ │ │ ├── collaborativeEditor.ts # CRDT three-way merge editor
│ │ │ │ ├── e2eEncryption.ts # RSA-4096 + AES-256-GCM
│ │ │ │ ├── faceMLService.ts # face-api.js model management
│ │ │ │ ├── multimodalAI.ts # Groq voice tone analysis
│ │ │ │ ├── videoCall.ts # WebRTC + STUN/TURN + SFU
│ │ │ │ ├── videoProctoring.ts # ML-based proctoring pipeline
│ │ │ │ ├── whiteboard.ts # Persistent whiteboard (MongoDB)
│ │ │ │ └── ... (81 total modules)
│ │ │ ├── middleware/ # Auth, RBAC, audit, security
│ │ │ ├── models/ # 7 Mongoose schemas
│ │ │ ├── routes/ # 38 Express route files
│ │ │ └── tests/ # Vitest + fast-check property-based tests
│ │ ├── models/face-api/ # ML model weights
│ │ └── Dockerfile # Multi-stage production build
│ │
│ └── web/ # Frontend Client
│ ├── src/
│ │ ├── components/ # Shadcn UI + custom components
│ │ ├── pages/ # Route pages (15+)
│ │ ├── hooks/ # Custom React hooks
│ │ ├── services/ # API client services
│ │ └── lib/ # Utilities
│ └── vercel.json # Deployment config
│
├── packages/
│ └── shared/ # Shared TypeScript types & interfaces
├── docs/ # Comprehensive documentation (7 files)
├── screenshots/ # Application screenshots
├── .github/workflows/ # CI/CD (GitHub Actions)
├── turbo.json # Turborepo build config
└── package.json # Root workspace configuration
| Tool | Version | Required |
|---|---|---|
| Node.js | 20+ | ✅ |
| npm | 10+ | ✅ |
| MongoDB | Atlas (free tier) or local | ✅ |
| Redis | Upstash (free) or local | Optional |
# Clone the repository
git clone https://github.com/theunstopabble/InterviewMinds.git
cd InterviewMinds
# Install all workspace dependencies
npm install --legacy-peer-deps
# Configure environment
cp apps/api/.env.example apps/api/.env
# Edit apps/api/.env with your API keys
# Start development (frontend + backend)
npm run dev| Service | URL | Description |
|---|---|---|
| Frontend | http://localhost:5173 | React app with hot reload |
| Backend API | http://localhost:8000 | Express REST + WebSocket |
| GraphQL | http://localhost:8000/graphql | Apollo Playground (dev only) |
| Health Check | http://localhost:8000/health | Service connectivity status |
| Metrics | http://localhost:8000/metrics | Prometheus exposition format |
# Run backend tests
cd apps/api && npm run test
# Run with coverage
cd apps/api && npm run test:coverage
# Watch mode
cd apps/api && npm run test:watchTest Suite: 39 property-based tests using Vitest 2.1 + fast-check 3.22
- 7 bug condition exploration tests (verify ML proctoring, CRDT merge, persistence, WebRTC)
- 32 preservation tests (verify no regressions across all 18 phases)
| Component | Platform | Status |
|---|---|---|
| Frontend | Vercel | ✅ Live |
| Backend | Render (Docker) | ✅ Live |
| Database | MongoDB Atlas (Free M0) | ✅ Connected |
| Cache/Queue | Upstash Redis (Free) | ✅ Connected |
Push to main → Typecheck (Backend + Frontend) → Test Suite → Build → Auto-deploy
The app requires only 5 keys to run. All other 45+ features gracefully degrade without their keys (no crash):
| Variable | Required | Free Tier |
|---|---|---|
MONGO_URI |
✅ | MongoDB Atlas Free (512MB) |
CLERK_PUBLISHABLE_KEY |
✅ | Clerk Free (10k MAU) |
CLERK_SECRET_KEY |
✅ | Clerk Free |
GROQ_API_KEY |
✅ | Groq Free (rate limited) |
JWT_SECRET |
✅ | Self-generated |
REDIS_URL |
Optional | Upstash Free (10k commands/day) |
CLOUDINARY_* |
Optional | Cloudinary Free (25GB) |
AZURE_SPEECH_* |
Optional | Azure Free (5hr/month) |
SENTRY_DSN |
Optional | Sentry Free (5k errors/month) |
Full list of 50 environment variables in
apps/api/.env.example
| Decision | Rationale |
|---|---|
| Groq over OpenAI | Sub-500ms latency, generous free tier, Llama 3.3-70b quality |
| CRDT Three-Way Merge | Preserves concurrent edits without data loss, no central coordination needed |
| face-api.js with fallback | ML inference when models available, content-derived analysis otherwise (never returns static values) |
| Circuit Breaker Pattern | Prevents cascading failures when Groq/Piston services go down |
| BullMQ with rate limiting | Reliable async jobs with pause-on-error to prevent Redis quota exhaustion |
| Clerk over custom auth | Enterprise-grade auth (OAuth, MFA, session management) without building from scratch |
| Turborepo monorepo | Shared types, parallel builds, intelligent caching across frontend + backend |
| Document | Description |
|---|---|
docs/API.md |
Complete REST + GraphQL API reference (150+ endpoints) |
docs/ARCHITECTURE.md |
System design, middleware pipeline, CRDT algorithm, ML pipeline |
docs/DB_SCHEMA.md |
7 MongoDB models with field-level documentation |
docs/DEPLOYMENT.md |
Production deployment (Vercel + Render + Docker) |
docs/EDGE_CASES.md |
Error handling, graceful degradation, failure modes |
docs/TECH_STACK.md |
Complete technology inventory (50+ packages) |
docs/WORKFLOW.md |
Development workflow, testing strategy, CI/CD |
- Fork the repository
- Create a feature branch:
git checkout -b feature/your-feature - Commit with conventional commits:
git commit -m 'feat: add feature' - Ensure
npm run typecheckpasses - Push and open a Pull Request — CI runs automatically
MIT License — see LICENSE for details.
InterviewMinds is a solo-developed enterprise platform showcasing full-stack engineering, AI/ML integration, real-time systems, and production-grade infrastructure.
Gautam Kumar — Full-Stack Developer & AI Engineer
Building enterprise-grade applications with React, Node.js, TypeScript, MongoDB, and AI/ML
| Project | Description | Link |
|---|---|---|
| Portfolio | Personal portfolio & developer profile | gautam-kr.vercel.app |
| SwadKart | Multi-vendor food delivery platform with AI chatbot | swadkart.vercel.app |
| Satark-AI | Deepfake detection & speaker verification platform | satark-deepfake.vercel.app |
| TexFolio | AI-powered LaTeX resume builder with RBAC | texfolio.vercel.app |