After Karpathy's US job exposure map went viral last week, I noticed nobody had built anything like it for Nigeria or any African country.
So I built it in a weekend.
What it does
aiexposure.com.ng scores 280 Nigerian occupations from 0-10 on AI automation exposure. Each job gets a score and a plain-English rationale explaining why.
The treemap visualization shows every job — area represents employment size, color represents AI exposure (navy = safe, coral = high risk).
The surprising finding
Nigeria's workforce averages 3.6/10 on AI exposure. The US averages 5.3/10.
Why the gap? Nigeria's economy runs on physical presence, cash, and interpersonal trust:
- Okada rider: 1/10 — AI can't navigate Lagos traffic
- POS agent: 3/10 — cash-based trust networks aren't digitizable
- Suya seller: 0/10 — no algorithm can read that fire
- Software developer: 9/10 — yes, us too
53% of Nigerian jobs score 0-3 (low risk). The informal sector that people often overlook is mathematically the most AI-proof workforce in the dataset.
But 18% of jobs score 7+. If your job lives on a screen, location doesn't matter — Lagos or London, same exposure.
Live AI scoring
The tool does something Karpathy's doesn't — you can type any job title and get an instant AI score. "Agege bread seller", "Bolt driver", "RCCG pastor" — anything.
This runs through a Netlify Function calling Claude Haiku at ~$0.0002 per request.
Tech stack
Intentionally minimal:
- Frontend: Single HTML file, Canvas-based treemap (no D3, no React, no frameworks)
- Backend: One Netlify Function for live scoring (Claude Haiku via Anthropic SDK)
- Data: Static JSON file with 280 pre-scored occupations
- Hosting: Netlify free tier
- Total API cost for scoring all 280 jobs: Under $2
The treemap uses a custom squarified layout algorithm rendered on HTML5 Canvas. No SVG, no libraries. The entire site is ~25KB excluding the data file.
How I built the dataset
- Started with Karpathy's 342 US occupations as a base
- Removed ~100 irrelevant ones (nuclear technicians, etc.)
- Added ~70 Nigeria-specific occupations (okada rider, POS agent, danfo driver, suya seller, generator technician, NYSC corp member, etc.)
- Estimated employment volumes using NBS sector data
- Scored each occupation using Claude with a structured rubric
- Generated rationales explaining each score
Open source
The entire project is open source: github.com/max-nvs/ai-jobs-ng
If you want to build something similar for your country, fork it. The scoring pipeline and visualization are reusable.
Try it
Check your job's AI exposure score: aiexposure.com.ng
Drop your job and score in the comments — I want to see who's safe and who's sweating.
Top comments (0)