AI Job Search Automation System

01

Problem

Applying to AI roles seriously means tracking dozens of opportunities across job boards, company sites, and recruiter conversations — each with its own stage, deadline, and follow-up timing. The failure mode isn't laziness, it's dropped threads.

02

Resume and job intelligence

LLM-based extraction turns the resume and each job description into structured fields — skills, requirements, seniority, constraints — so matching is a structured comparison rather than intuition. Roles are ranked by fit before any time is spent on them.

03

Recruiter workflow with human gates

n8n workflows classify incoming email context, build a structured prompt, and generate a draft reply or follow-up. Nothing sends without explicit approval — the same pattern I'd deploy for any business email automation: AI drafts, humans decide.

• Intent classification on incoming messages
• Context-aware draft generation
• Mandatory human review gate before sending
• Outcome logging for traceability

04

Application tracking

A SQLite-backed tracker records every role, its stage, dates, and artifacts. The CLI copilot surfaces what needs action today — follow-ups due, deadlines approaching — instead of relying on memory.