HackCanadaApp is an AI-powered interview preparation platform with a React frontend and Python backends. It combines interview discovery, personalized prep context, mock interviews, and multimodal feedback (speech, stress/body-language cues, and answer quality).

• Frontend: Vite + React app in Behaviourly-frontend/Behviourly-frontend.
• FastAPI backend (hackcanada-backend/main.py, default port 8001): mock interview generation, video analysis, transcription, answer analysis, reporting, and session state.
• Flask/Auth backend (hackcanada-backend/Auth.py, default port 8000): Auth0 login flow, interview records, interview context + question retrieval/generation.
• Scanner worker (hackcanada-backend/Scanner.py): optional Gmail scanner that detects interview emails, stores interviews, and can send WhatsApp notifications.

1. User signs in via Auth0 (Google social connection).
2. Dashboard lists detected interviews (or user can create one manually).
3. User opens an interview prep kit (/interview-context) with company/role briefing.
4. User starts mock interview (/interview) with generated/fallback questions.
5. During each answer, frontend records video and submits it for AI feedback.
6. Backend returns transcript + analysis (Gemini, optional Presage physiology) and aggregates data for final report.

• React 19
• React Router
• Vite
• CSS modules/files per page/component

• FastAPI + Uvicorn
• Flask + SQLAlchemy
• Google Gemini (google-genai)
• OpenAI Whisper fallback path for transcription (if configured)
• Presage Physiology API integration (video upload + vitals retrieval)
• Auth0 OAuth + Management API

.
├── README.md
├── docs/
│   └── PRESAGE_REALTIME_PLAN.md
├── hackcanada-backend/
│   ├── main.py                 # FastAPI interview + analysis API
│   ├── Auth.py                 # Flask Auth0 + interview CRUD API
│   ├── Scanner.py              # Optional background Gmail scanner
│   ├── interview_context.py    # FastAPI router for interview context generation
│   ├── requirements.txt
│   └── users.db                # Local SQLite DB (development)
├── Behaviourly-frontend/
│   └── Behviourly-frontend/
│       ├── src/
│       ├── package.json
│       └── vite.config.js
└── package.json                # Root scripts to run frontend + FastAPI together

• Node.js 18+
• npm
• Python 3.10+
• Auth0 tenant/apps (for login flow)
• API keys depending on features:
  • GEMINI_API_KEY (required for most AI endpoints)
  • OPENAI_API_KEY (optional transcription path)
  • PRESAGE_API_KEY (optional physiology analysis)

Create .env at the repo root (or in hackcanada-backend/ for backend-only local use).

# AI
GEMINI_API_KEY=...
OPENAI_API_KEY=...
PRESAGE_API_KEY=...
ELEVENLABS_API_KEY=...
ELEVENLABS_VOICE_ID=...
ELEVENLABS_MODEL_ID=eleven_multilingual_v2

# Auth0 / Flask auth service
AUTH0_DOMAIN=...
AUTH0_CLIENT_ID=...
AUTH0_CLIENT_SECRET=...
AUTH0_M2M_CLIENT_ID=...
AUTH0_M2M_CLIENT_SECRET=...
FLASK_SECRET_KEY=...
REDIRECT_URI=http://localhost:8000/callback
FRONTEND_URL=http://localhost:5173
DATABASE_URL=sqlite:///users.db

# Optional scanner + Twilio
SCAN_INTERVAL=1800
TWILIO_ACCOUNT_SID=...
TWILIO_AUTH_TOKEN=...
TWILIO_WHATSAPP_NUMBER=whatsapp:+14155238886
MY_PERSONAL_PHONE=whatsapp:+1XXXXXXXXXX

Create Behaviourly-frontend/Behviourly-frontend/.env if you need overrides:

VITE_API_BASE_URL=http://127.0.0.1:8001
VITE_AUTH_API_BASE_URL=http://localhost:8000
VITE_ELEVENLABS_VOICE_ID= # optional override; backend ELEVENLABS_VOICE_ID is default
VITE_ELEVENLABS_MODEL_ID= # optional override; backend ELEVENLABS_MODEL_ID is default

npm install
npm install --prefix Behaviourly-frontend/Behviourly-frontend

python -m venv .venv
source .venv/bin/activate
pip install -r hackcanada-backend/requirements.txt

From repo root:

npm start

This runs:

• FastAPI backend on http://0.0.0.0:8001
• Vite frontend on http://localhost:5173

cd hackcanada-backend
uvicorn main:app --reload --host 0.0.0.0 --port 8001

cd hackcanada-backend
python Auth.py

npm run dev --prefix Behaviourly-frontend/Behviourly-frontend

• GET /health - liveness check
• GET /health/ready - readiness check for Gemini-dependent paths
• POST /mock-interview - generate interview questions
• POST /gemini/analyze-video - analyze uploaded answer video
• POST /transcribe - transcript extraction from uploaded media
• POST /tts - ElevenLabs text-to-speech proxy (returns audio)
• POST /analyze-answer - evaluate answer quality for a question
• POST /presage/analyze - optional physiology analysis from video
• POST /report - generate final interview report
• POST /reset-session - reset in-memory scoring/session
• WebSocket /ws/live - live signals/events stream

• GET /login, GET /callback, GET /logout - authentication flow
• GET /me - current user info from session
• GET /interviews - list interview records for user
• GET /interviews/<id>/questions - fetch saved questions
• POST /interviews/<id>/generate-questions - generate and persist interview questions
• GET /interviews/<id>/context - fetch/generate interview prep context

hackcanada-backend/Scanner.py can be run in one-shot or loop mode to detect interview emails in Gmail.

cd hackcanada-backend
python Scanner.py
# or
python Scanner.py --loop

It relies on Auth0-stored Google tokens and persists interview records/questions in the local database.

• The frontend folder is intentionally named Behviourly-frontend (typo preserved in existing scripts).
• hackcanada-backend/users.db is committed and used for local development.
• Presage Python SDK is currently commented out in requirements due dependency conflicts; REST API flow is used/planned in this repo.
• Some frontend endpoints are hardcoded to http://localhost:8000 for auth-related routes.

• Presage integration planning: docs/PRESAGE_REALTIME_PLAN.md

• If /interview shows backend or AI not ready:
  • Ensure FastAPI is running on :8001.
  • Ensure GEMINI_API_KEY is set.
• If login loops or callback fails:
  • Verify Auth0 callback/logout URLs match local ports.
  • Ensure Flask/Auth service is running on :8000.
• If interview dashboard is empty:
  • Log in first (/login) and verify session cookies are enabled.
  • Ensure interview records exist in the local DB for your logged-in user.