Interview Copilot — your AI wingman for job interviews

An always-on voice copilot that listens to your interview and feeds you perfect, CV-grounded answers in real time.

Built for the Nebius x NVIDIA Global AI Hackathon (Personal AI track).

How it works

1. Click Start Session and grant mic access
2. The copilot transcribes the interviewer's speech live (Web Speech API)
3. When it detects a question, NVIDIA Nemotron (via Nebius Token Factory) generates a concise answer grounded in your CV
4. The answer appears instantly — read it, or have it whispered via TTS
5. Hit Deep Dive on tough questions for a detailed STAR-format answer

Tech

• LLM: nvidia/Nemotron-3_5-Lightning (real-time answers) + nvidia/nemotron-3-super-120b-a12b (deep dives) via Nebius Token Factory (NEBIUS_API_KEY)
• Backend: Python FastAPI, OpenAI-compatible client
• Frontend: Vite + vanilla JS, Web Speech API (recognition + synthesis)
• Fallback: FastRouter (Claude Opus) if Nebius key is absent — dev only

Quickstart

# backend
cd backend
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
export NEBIUS_API_KEY=your-key-here
uvicorn app.main:app --port 8771

# frontend (served by backend in prod; dev:)
cd frontend
python -m http.server 8772
# open http://localhost:8772

Layout

• backend/app/main.py — FastAPI app, SSE answer streaming
• backend/app/nemotron.py — Nebius Token Factory client (OpenAI SDK)
• backend/app/cv_context.py — CV grounding context
• backend/app/questions.py — question detection heuristics
• frontend/index.html — copilot UI

License

MIT