# 🤖 Interview Bot

AI-powered interviewer bot with video call interface and calendar scheduling, built on Cloudflare Workers.

## Features

- **📅 Calendar Scheduling** - Visual calendar with drag-and-click scheduling
- **📹 Video Call Interface** - Simulated video call with AI avatar
- **🧠 AI Interviewer** - Streaming responses via Workers AI + AI Gateway
- **💬 Real-time Chat** - Server-sent events for smooth streaming
- **📱 Responsive UI** - Works on desktop and mobile

## Architecture

```
┌─────────────────────────────────────────────────────────┐
│                    Cloudflare Edge                       │
│  ┌─────────────┐  ┌──────────────┐  ┌────────────────┐  │
│  │   Worker    │──│  AI Gateway  │──│   Workers AI   │  │
│  │  (index.ts) │  │  (optional)  │  │ (Llama 3.1 8B) │  │
│  └──────┬──────┘  └──────────────┘  └────────────────┘  │
│         │                                                │
│  ┌──────┴──────┐                                        │
│  │  KV Store   │ ← Interview data                       │
│  └─────────────┘                                        │
└─────────────────────────────────────────────────────────┘
```

## Quick Start

### 1. Install dependencies

```bash
npm install
```

### 2. Create KV namespaces

```bash
# Create production namespace
npx wrangler kv namespace create INTERVIEWS

# Create preview namespace for local dev
npx wrangler kv namespace create INTERVIEWS --preview
```

### 3. Update wrangler.toml

Replace the placeholder IDs with the ones from the previous step:

```toml
[[kv_namespaces]]
binding = "INTERVIEWS"
id = "<YOUR_PRODUCTION_ID>"
preview_id = "<YOUR_PREVIEW_ID>"
```

### 4. (Optional) Configure AI Gateway

For analytics, caching, and rate limiting:

1. Create an AI Gateway in the Cloudflare dashboard
2. Uncomment and update the AI Gateway config in `wrangler.toml`:

```toml
[ai.gateway]
id = "your-gateway-id"
```

### 5. Run locally

```bash
npm run dev
```

Open http://localhost:8787

### 6. Deploy to production

```bash
npm run deploy
```

## API Endpoints

| Method | Endpoint | Description |
|--------|----------|-------------|
| GET | `/api/interviews` | List all interviews |
| POST | `/api/interviews` | Create new interview |
| PUT | `/api/interviews/:id` | Update interview |
| DELETE | `/api/interviews/:id` | Delete interview |
| POST | `/api/chat` | Stream AI chat response |

## Interview Data Structure

```typescript
interface Interview {
  id: string;
  candidateName: string;
  email: string;
  scheduledAt: string; // ISO 8601
  duration: number;    // minutes
  status: 'scheduled' | 'in-progress' | 'completed' | 'cancelled';
  notes: string[];
  transcript: string[];
}
```

## Customization

### Change AI Model

In `src/index.ts`, update the model in `handleChat()`:

```typescript
const stream = await env.AI.run('@cf/meta/llama-3.3-70b-instruct-fp8-fast', {
  // ...options
});
```

Available models: https://developers.cloudflare.com/workers-ai/models/

### Modify Interview Prompts

Edit the `SYSTEM_PROMPT` constant in `src/index.ts` to customize:
- Interview style and tone
- Question categories
- Evaluation criteria

### Add Real Video (WebRTC)

To add actual video calls using Cloudflare Realtime:

1. Enable Realtime SFU in your Cloudflare dashboard
2. Add the RealtimeKit SDK to the frontend
3. Replace the simulated video with real WebRTC streams

See: https://developers.cloudflare.com/realtime/

## Tech Stack

- **Runtime**: Cloudflare Workers
- **AI**: Workers AI with Llama 3.1 8B Instruct
- **Storage**: Cloudflare KV
- **Streaming**: Server-Sent Events (SSE)
- **Frontend**: Vanilla JS with inline styles

## Cost Estimation

| Resource | Free Tier | Paid |
|----------|-----------|------|
| Workers requests | 100k/day | $0.30/million |
| Workers AI | 10k neurons/day | $0.011/1k neurons |
| KV reads | 100k/day | $0.50/million |
| KV writes | 1k/day | $5.00/million |

## License

MIT