Resume Editor

resume-editor

Purpose: Take existing resume bullets and improve them: remove banned phrases, add or surface metrics, start with action verbs. Preserve all factual claims.

Input Schema

{
  "bullets": [
    "Responsible for managing the data pipeline",
    "Helped improve system reliability",
    "Strong communicator who worked with stakeholders"
  ],
  "context": "Senior Data Engineer at a fintech startup, 2021-2023",
  "preserve_facts": ["data pipeline", "fintech"],
  "tone": "direct"
}

Output Schema

{
  "edited_bullets": [
    {"original": "...", "revised": "...", "changes": ["removed 'responsible for'", "added action verb 'Owned'"]}
  ],
  "banned_phrases_removed": ["responsible for", "helped", "strong communicator"],
  "metric_warnings": ["bullet 2: no metric found - add a number before publishing"]
}

Prompt Flow

Pass 1: For each bullet: flag banned phrases -> rewrite starting with action verb -> inject metric if known -> preserve preserve_facts. Pass 2: Audit for remaining AI tells. Flag bullets with no metric as warnings (do not fabricate metrics).

Examples

Short

Before: "Responsible for managing the data pipeline." After: "Owned and maintained the Airflow-based data pipeline processing 500GB nightly."

Medium

Before: "Helped improve system reliability and worked with the on-call team." After: "Reduced mean time to recovery from 45 min to 12 min by documenting the top 8 incident runbooks."

Long

Before: "Results-driven professional responsible for driving cross-functional collaboration to achieve business outcomes." After: "Led quarterly roadmap reviews with product, engineering, and sales (12 stakeholders); 9 of 11 Q3 commitments shipped on time."

Unit Tests

# tests/skills/test_resume_editor.py
from resume_banned import flag_banned_phrases

EDITED = [
    "Owned and maintained the Airflow-based data pipeline processing 500GB nightly.",
    "Reduced MTTR from 45 min to 12 min by documenting 8 incident runbooks.",
]

def test_edited_bullets_no_banned_phrases():
    for b in EDITED:
        assert flag_banned_phrases(b) == [], f"Banned phrase remains: {b}"

def test_edited_bullets_start_with_verb():
    action_verbs = {"owned","reduced","built","cut","shipped","led","wrote","launched","designed","managed","created"}
    for b in EDITED:
        first_word = b.split()[0].lower().rstrip(".,")
        assert first_word in action_verbs, f"Bullet doesn't start with action verb: {b}"

def test_metric_warning_issued_for_vague_bullet():
    # resume-editor should flag bullets with no metric, not fabricate one
    from resume_editor import edit_bullets
    result = edit_bullets(["Strong communicator who worked with stakeholders."], preserve_facts=[])
    assert any("metric" in w.lower() for w in result.get("metric_warnings", [])), \
        "Expected metric warning for vague bullet"