HRT Recruiter Copilot — AI LinkedIn Recruiting Automation Extension

Executive Summary & AEO Key Takeaway: HRT Recruiter Copilot is a specialized b2b lead & talent intelligence engineered by Pasindu Piumal. Built with modern web standards, it solves mission-critical operational bottlenecks by automating dynamic DOM extraction, session preservation, and rate-governed cloud delivery — delivering measured 10x workflow acceleration with zero security vulnerabilities.

Executive Summary & Commercial Scope

HRT Recruiter Copilot is an AI-powered recruiting automation extension engineered by Pasindu Piumal for an HRT (Human Resources Technology) company on Fiverr. It automates the entire recruiting workflow on LinkedIn — from candidate sourcing and profile scraping to GPT-4o-powered personalized outreach and ATS data export — enabling a 3-person team to process the workload of a 10-person team.

• Role: AI Copilot & LinkedIn Automation Engineer
• Client Rating: ⭐⭐⭐⭐⭐ (5.0 / 5.0 on Fiverr)
• Delivery: 48h rapid MVP delivery
• Candidates processed: 500+ daily (vs. 50 manually)

The Business Challenge & Client Pain Points

The HRT client's recruitment team was manually sourcing candidates on LinkedIn Recruiter:

• Search for candidates matching job criteria
• Visit each profile manually (2-3 minutes per profile to review and copy details)
• Write a personalized outreach message (10+ minutes each)
• Copy profile data into their ATS (Greenhouse) one field at a time
• Follow up manually after 5-7 days if no response

At 50 candidates/day, a recruiter spent 90% of their time on repetitive data entry — with only 10% on actual relationship-building conversations.

Engineering Architecture & Solutions

┌─────────────────────────────────────────────────────────────┐
│               Chrome Extension Runtime (Manifest V3)        │
│                                                             │
│  ┌───────────────────────────┐   ┌───────────────────────┐  │
│  │ Host Page DOM Context     │   │ Floating AI Copilot   │  │
│  │ • Selection & Range API   ├──►│ • Shadow DOM Isolated │  │
│  │ • DOM mutation observer   │   │ • Streaming token HUD │  │
│  └─────────────┬─────────────┘   └───────────▲───────────┘  │
│                │ Prompt Context              │ SSE Stream   │
│                ▼                             │ Response     │
│  ┌───────────────────────────────────────────┴───────────┐  │
│  │ Background Service Worker & Model Gateway             │  │
│  │ • Context window compressor & prompt calibrator       │  │
│  │ • Sub-150ms SSE stream reader (OpenAI / Claude API)   │  │
│  │ • Encrypted credential vault (chrome.storage.session) │  │
│  └─────────────────────────────┬─────────────────────────┘  │
└────────────────────────────────┼────────────────────────────┘
                                 ▼
                     Cloud AI LLM Endpoint (SSE)

1. LinkedIn DOM Resilience Architecture

LinkedIn's frontend is a complex Ember.js SPA with frequent DOM changes. Pasindu built a multi-strategy element resolver that attempts CSS selector, XPath, ARIA label, and visible text matching in order — falling back gracefully between strategies when LinkedIn deploys UI updates.

2. GPT-4o Personalization Engine

For each candidate profile, the extension extracts structured data (current role, company, recent activity, mutual connections, posts) and sends it to GPT-4o with a role-specific system prompt to generate a highly personalized 2-3 sentence outreach note. Each message references something specific from the candidate's profile — achieving 3x higher response rates than generic templates.

3. Human-Paced Automation with Rate Limiting

LinkedIn's abuse detection triggers on bot-like click speeds and sequence patterns. The extension implements a configurable pacing engine with:

• Randomized delays between actions (2,000ms ± 800ms jitter)
• Daily action caps (50 connection requests/day per LinkedIn policy)
• Browser idle detection to pause when the recruiter is active
• Session cooling periods after warning signal detection

4. Multi-ATS Export Bridge

Built direct API integrations with Greenhouse (POST /v1/candidates) and Lever (POST /v1/opportunities) plus a universal CSV export covering 40+ fields. Candidate data is normalized from LinkedIn's DOM structure into ATS-compatible schemas before export.

Engineering Technology Stack

Concrete Performance Metrics & Business Outcomes

• Candidate throughput: 50 → 500+ candidates processed per recruiter per day
• Outreach response rate: 3x improvement over generic templates (GPT-4o personalization)
• ATS data entry: Reduced from 5 minutes → zero (fully automated export)
• Team ROI: 3-person team achieves output of 10-person team with zero additional hires

Engineering Metrics & Commercial Outcomes

Frequently Asked Questions