generate-emerging-job-titles

name	generate-emerging-job-titles
description	Generate real-world job title aliases for O*NET occupations and merge them into ai_resilience_scores.csv. Use when adding a new cluster or when O*NET sample titles don't reflect how jobs are actually listed.

Generate Emerging Job Titles Skill

Discovers real-world job titles that map to existing ONET codes but don't appear in ONET's Sample Job Titles. Writes to data/emerging_roles/emerging_job_titles.csv and syncs the Emerging Job Titles column in data/output/ai_resilience_scores.csv.

Why this matters: O*NET titles are bureaucratic ("Market Research Analysts and Marketing Specialists"). Job seekers search for "Social Media Manager" or "Content Strategist." These aliases power site search and career page discoverability.

---

Commands

# Generate + merge for a cluster (recommended — run after career-clusters skill)
python3 scripts/generate_emerging_job_titles.py --cluster office-admin

# Single occupation
python3 scripts/generate_emerging_job_titles.py --code 13-1161.00

# Just re-sync CSV → scores (no API calls — use after manual edits to the CSV)
python3 scripts/generate_emerging_job_titles.py --merge-only

# All occupations (slow — use per cluster)
python3 scripts/generate_emerging_job_titles.py --all

Requires ANTHROPIC_API_KEY. Skips occupations that already have entries in emerging_job_titles.csv.

---

What the script does

1. For each occupation, calls Claude with the O*NET title + existing Sample Job Titles
2. Claude returns 2–5 real-world aliases not already in the sample titles list
3. Appends new rows to emerging_job_titles.csv
4. Merges all entries into the Emerging Job Titles column in ai_resilience_scores.csv

---

QA after generation

Review the new rows in emerging_job_titles.csv. For each title check:

1. Real postings exist — search LinkedIn Jobs for the title. Should have >50 postings in past 30 days. If not, delete the row.
2. Title is specific — "AI Specialist" is too generic. "AI Content Strategist" is acceptable.
3. Mapping is accurate — the title should genuinely describe work done under that O*NET code, not just sound related.

To delete a bad row: edit emerging_job_titles.csv directly, then run --merge-only to sync.

---

Manual additions

To add a title manually (e.g. from your own research or the emerging roles research guide):

1. Add a row to data/emerging_roles/emerging_job_titles.csv:

   onet_code,job_title,notes
   13-1161.00,Social Media Manager,Manages brand social presence; closer to Marketing Manager scope
   

2. Run: python3 scripts/generate_emerging_job_titles.py --merge-only

---

Cache behavior

The script skips any onet_code that already has at least one entry in emerging_job_titles.csv. To regenerate for a code, delete its rows from the CSV first, then re-run.

---

Output

data/emerging_roles/emerging_job_titles.csv schema:

Field	Description
onet_code	O*NET code the title maps to
job_title	Real-world title as it appears in job postings
notes	Why it maps here; what differs from the O*NET title

---

After this skill

python3 scripts/generate_next_steps.py --cluster <id>
python3 scripts/adjacent_roles.py --cluster <id>

Also: the Emerging Job Titles column in ai_resilience_scores.csv is now populated and available for site search and career page generation.