ai-talent-recruiter

# AI Talent Recruiter

A comprehensive skill for AI/ML talent discovery, profiling, and personalized outreach.

## Overview

This skill enables the complete talent recruitment pipeline:
1. **Search**: Generate optimized queries and discover candidates via web search
2. **Extract**: Scrape profile pages and extract structured candidate information
3. **Identify**: Recognize Chinese candidates using surname and institution matching
4. **Classify**: Categorize candidates by type (PhD/PostDoc/Professor/Industry)
5. **Deduplicate**: Remove duplicate candidates using 7-level fingerprinting
6. **Standardize**: Map research fields to 22 standardized categories
7. **Generate**: Create personalized outreach emails using field-specific templates

## Scraping Approach Selection

This skill supports two scraping approaches:

| Approach | Best For | Cost | Success Rate |
|----------|----------|------|--------------|
| **BrightData MCP** (Recommended) | LinkedIn, Twitter, high-anti-scraping sites | Paid | High (99%+) |
| **Python Scraping** | Academic sites (.edu), personal homepages | Free | Medium (70-85%) |

### Selection Guidelines

- **Use Python scraping** for academic websites (.edu, .github.io) - **FREE**
- **Use BrightData MCP** for LinkedIn and sites requiring login - **PAID**
- See `references/python-scraping-guide.md` for detailed implementation

## Prerequisites

### Option A: BrightData MCP (Paid)

This approach requires the BrightData MCP server to be configured in Claude Code.

1. Get your API token from [BrightData](https://brightdata.com)
2. Add the MCP server using Claude CLI:

```bash
claude mcp add --transport sse --scope user brightdata "https://mcp.brightdata.com/sse?token=<your-api-token>"
```

Replace `<your-api-token>` with your actual BrightData API key.

3. Verify the MCP server is connected by checking available tools:
   - `mcp__brightdata__search_engine` - For web searches
   - `mcp__brightdata__scrape_a...

# AI Talent Recruiter Skill

A comprehensive Claude Code skill for AI/ML talent discovery, profiling, and personalized outreach.

## Features

- **Multi-source Search**: Search candidates via web search, academic sites, and professional networks
- **Dual Scraping Approach**: Support both BrightData MCP (paid) and Python (free) scraping
- **Chinese Candidate Identification**: Multi-dimensional scoring using surnames and institutions
- **Smart Classification**: Categorize candidates as PhD/PostDoc/Professor/Industry
- **7-Level Deduplication**: Remove duplicate candidates using email, Scholar, LinkedIn, etc.
- **22 Standardized Fields**: Map research directions to standard categories
- **Personalized Emails**: Generate customized outreach using field-specific templates

## Installation

### Step 1: Copy to Claude Skills Directory

```bash
# Copy the entire ai-talent-search folder to your Claude skills directory
cp -r ai-talent-search ~/.claude/skills/
```

### Step 2: Configure Prerequisites

**Option A: BrightData MCP (Paid, Recommended)**

```bash
claude mcp add --transport sse --scope user brightdata "https://mcp.brightdata.com/sse?token=<your-api-token>"
```

**Option B: Python Scraping (Free)**

```bash
pip install requests beautifulsoup4 httpx python-dotenv
```

### Step 3: Verify Installation

In Claude Code, type `/ai-talent-recruiter` to invoke the skill.

## Quick Start

### Example 1: Search RL PhD Students

```
User: Use the ai-talent-recruiter skill to find PhD students working on reinforcement learning at top universities
```

Claude will:
1. Generate search queries for RL PhD students
2. Execute searches and collect URLs
3. Scrape profiles (using BrightData or Python)
4. Extract and classify candidates
5. Generate personalized emails

### Example 2: Lab-Targeted Search

```
User: Find all PhD students at Stanford AI Lab working on multimodal learning
```

### Example 3: Chinese Candidate Disc...