LLM Interview Simulator

An interview simulation platform that uses Large Language Models (LLMs) to conduct realistic technical and behavioural interviews and provide detailed feedback. https://vdhaya-interview-simulator.com

Project Structure

• Backend: FastAPI-based Python server (interview API, evaluation, health checks). Lives in backend/.
• Frontend: React application (interview UI, analytics). Lives in frontend/.
• Documentation: See docs/ for architecture, AI tools, and research notes.

Features

• Real-time interview simulation with LLM-powered interviewer
• Comprehensive feedback scoring and analysis
• Analytics dashboard for performance tracking
• Support for various technical interview formats
• CORS, health checks, Azure and Docker deployment support

Quick Start

Backend

# From repo root
pip install -r requirements.txt
cd backend
# Copy .env.example to .env, set DATABASE_URL (and optionally SECRET_KEY, AWS_*)
# alembic upgrade head  # if using DB
uvicorn app.main:app --host 0.0.0.0 --port 8000

• API: http://localhost:8000
• Docs: http://localhost:8000/docs
• Health: http://localhost:8000/health

Frontend

cd frontend
npm install
npm run dev

Backend API (summary)

• GET /, GET /health – health and status
• GET /api/interview/ – sample interview questions
• POST /api/interview/evaluate – evaluate an answer
• GET /api/interview/types – available interview types

Docker

From project root:

docker build -f backend/Dockerfile -t llm-interview-simulator .
docker run -p 8000:8000 llm-interview-simulator

Elastic Beanstalk (API + React from one URL)

From project root, build the frontend into backend/app/static and deploy the backend:

./scripts/deploy-eb-fullstack.sh

The UI uses same-origin /api/... calls (VITE_API_BASE_URL empty in that build).

Set DATABASE_URL in the EB environment (Configuration → Software → Environment properties) to your RDS PostgreSQL URL. Each deploy runs alembic upgrade head in the container entrypoint.

Migrations stuck (tables missing but older tables exist): If the DB was created without Alembic (or alembic_version is empty while users / interview_sessions already exist), upgrade head from scratch fails with “already exists”. Fix once from your laptop (same DATABASE_URL as EB):

cd backend
export DATABASE_URL='postgresql://...'   # RDS URL from EB
alembic stamp b2c3d4e5f6a7   # last revision before daily_visits + calendar; adjust if your schema differs
alembic upgrade head

Then confirm user_daily_visits and user_calendar_events exist. Redeploy afterward so future upgrades apply normally.

Azure

• App Service: Connect repo, set startup command python startup.py, set PORT=8000 and other env vars.
• Container: Build/push image to ACR, then deploy container (see Azure docs).

Configuration

Backend uses a .env file in backend/ (e.g. PORT, ENVIRONMENT, DATABASE_URL, SECRET_KEY, AWS_*). Copy backend/.env.example to backend/.env.

Testing (backend)

pip install pytest pytest-asyncio
cd backend && pytest tests/

Troubleshooting

• Port 8000 in use: lsof -ti:8000 | xargs kill -9
• Import errors: Run from repo root or backend/; ensure deps: pip install -r requirements.txt
• Use http://localhost:8000/docs for interactive API testing

Documentation

See the docs/ directory for detailed documentation.

License

This project is licensed under the MIT License - see the LICENSE file for details.