While building Ghost-Recruiter, I faced a major dilemma: How much should we automate?
On one hand, AI (n8n + Gemini) is incredibly fast at parsing 500+ resumes and finding technical matches. On the other hand, a machine might miss that "special spark" or cultural fit that only a human recruiter can feel.
My approach was to build a "Human-in-the-loop" system where AI handles the data noise, but humans make the final call.
Recruiters spend 70% of their time on manual resume screening and repetitive follow-ups. Ghost Recruiter is an autonomous AI ecosystem that replicates the operations of an entire HR department by understanding context, verifying data, and continuously learning from its own decisions.
⚙️ Technical Stack
Engine: n8n (Enterprise Workflow Automation)
Cognitive Services: Google Gemini 3.1 Flash (LLM)
Data Hub: Google Sheets (Dynamic DB & Job Requirements)
Communication: Telegram Bot API & Gmail API
