Smart Ai Interviewer

Adaptive Voice-Based Interview Simulation & Weakness Mapping Platform

Smart Ai Interviewer is an AI-driven interview simulation platform that uses voice interaction to provide realistic, adaptive mock interviews. Unlike conventional mock-interview apps (mostly text-only or static), our system speaks questions aloud (using text-to-speech) and listens to spoken answers (using speech-to-text). It dynamically adapts each next question based on previous answers, employing large language models (LLMs) and NLP to understand the context, analysing the given answer for Clarity and Confidence and also Scoring based on that. After the interview, it generates a structured report highlighting strengths and weaknesses (weakness mapping), along with actionable feedback. Early internal tests show that users find the voice-based interaction more engaging and realistic than text-only quizzes.

Tech Stack

Layer	Technology
Frontend	Next.js 15, TypeScript, Tailwind CSS, Framer Motion
Backend	Node.js, NestJS (7 microservices)
Database	PostgreSQL 16 + Prisma ORM
Cache/Queue	Redis 7 + BullMQ
AI	Gemini (free tier), Local Whisper (STT), edge-tts (TTS)
Infra	Docker Compose, Turborepo monorepo

Project Structure

voxhire-ai/
├── apps/
│   ├── web/                          # Next.js frontend
│   └── services/
│       ├── api-gateway/              # Entry point, rate limiting, routing
│       ├── auth-service/             # JWT auth, Redis session cache
│       ├── interview-service/        # Core orchestrator (adaptive logic)
│       ├── voice-service/            # edge-tts + Whisper STT
│       ├── scoring-service/          # AI evaluation + aggregation
│       ├── analytics-service/        # Reports, weakness mapping, history
│       └── storage-service/          # File storage (audio, reports)
├── packages/
│   ├── shared-types/                 # TypeScript types & DTOs
│   ├── config/                       # Centralized env config (zod)
│   ├── db/                           # Prisma schema & client
│   └── ai-provider/                  # Abstracted AI interface (Gemini)
├── docker-compose.yml                # PostgreSQL + Redis
├── turbo.json                        # Turborepo config
└── docs/                             # Product & engineering docs

Quick Start

Prerequisites

• Node.js 18+
• Docker & Docker Compose
• Python 3.8+ (for Whisper & edge-tts)

1. Clone & Install

git clone <repo-url>
cd voxhire-ai
cp .env.example .env    # Edit with your Gemini API key
npm install

2. Start Infrastructure

docker-compose up -d    # Starts PostgreSQL + Redis

3. Setup Database

cd packages/db
npx prisma generate
npx prisma db push

cd ..

4. Install Python Dependencies

pip install edge-tts openai-whisper

5. Run Development

npm run dev             # Starts all services + frontend

Individual Services

npm run dev:web         # Frontend only (localhost:3000)
npm run dev:gateway     # API Gateway only (localhost:4000)
npm run dev:services    # All backend services

Service Ports

Service	Port
Frontend	3000
API Gateway	4000
Auth Service	4001
Interview Service	4002
Voice Service	4003
Scoring Service	4004
Analytics Service	4005
Storage Service	4006
PostgreSQL	5432
Redis	6379

Environment Variables

See .env.example for all required variables.

License

Private project.