I treat job applications as a pipeline, not a resume problem

Description and use

A job‑application pipeline that automatically sources, scores, drafts, and queues applications, sending them only after human approval.

Problem

Job applications often fail before reaching a human due to timing, source quality, and generic content rather than resume quality.

Outcome

The pipeline reduces wasted applications, ensures timely submissions, and keeps human judgment as the final gate, leading to higher quality outreach and better candidate experience.

Workflow

• Schedule nightly job to scrape multiple job boards and feeds
• Log source yield and adjust targeting based on zero‑result days
• Score each role against CV and truth file, filtering out unsupported claims
• Draft cover letter constrained to truth file facts
• Create interactive Discord/Telegram card with role, fit score, and draft preview
• Human approves, edits, or rejects the draft; rejections logged for future training
• Send approved applications at scheduled time
• Enforce caps on candidates per night and sends per month
• Persist state and counters locally in the user’s repo
• Use failover LLM providers to avoid rate‑limit stalls

Evidence

The whole thing is a Python package plus a repo with a scheduler.

Source

Drafted applications land as interactive cards on Discord (or Telegram).

Source