affectvision-ai

Static multi-tool prototype with separate modules for interview preparation and interview process monitoring.

Current modules

• Home tools selector: index.html
• Coaching and Preparation (single-screen, Chrome-only): tools/coaching/index.html
• Interview Monitoring Console (organization view): tools/interview/index.html
• Candidate Session Screen (for monitoring flow): tools/interview/interviewee.html

Run locally

1. Start a static server from this folder.
2. Open the served URL in a modern browser.
3. From home, select either:
   • Coaching and Preparation for personal practice (Google Chrome required), or
   • Interview Monitoring for interviewer-driven sessions.
4. For monitoring flow, use Open Candidate Session Screen to open the candidate view in a new tab/window.
5. Allow camera permission when prompted on the active camera screen.

Example servers:

• npx serve .
• python -m http.server 8000

Do not open with file://; model loading and module imports require HTTP.

Project structure

• index.html and home.css: tool selection home page
• tools/coaching/: single-screen coaching and preparation module
• tools/interview/: interviewer monitoring and candidate session screens
• shared/interview-session-engine.js: reusable camera + face-api session engine
• shared/interview-channel.js: cross-tab sync for live question updates
• models/: local face-api.js weights
• docs/: product and roadmap documents

Coaching flow notes

• Users complete a survey first (interview type, topic, difficulty, question count).
• Questions are then asked one by one.
• Users can answer by text or audio (speech-to-text in Chrome).
• Each response is evaluated per question.
• Final report combines response quality and behavioral signals.
• Chrome AI is used when available; heuristic fallback is applied otherwise.

Deploy to GitHub Pages

1. Push this repository to GitHub on branch main.
2. In GitHub repo settings, open Pages.
3. Set Source to GitHub Actions.
4. Commit/push any change (or run the workflow manually) to trigger deployment.
5. Open the published URL shown in the Deploy static site to GitHub Pages workflow run.

Notes:

• Workflow file: .github/workflows/deploy-pages.yml
• .nojekyll is included to avoid Jekyll processing issues.
• Detailed deployment/troubleshooting: DEPLOYMENT.md