Three engines.
Independently callable.
Five-minute integration.

UNMAPPED is open skills-intelligence infrastructure for LMICs. You don't need to adopt our UI — mount any of the three engines in your own product. Every response is taxonomy-coded (ESCO / ISCO-08 / O*NET) and cites its data source.

ENGINE 01

Skills Signal

Story → portable, taxonomy-coded profile.

ENGINE 02

Risk Lens

LMIC-calibrated automation risk + Wittgenstein 2035.

ENGINE 03

Opportunity Match

Skills → ranked jobs + econometric signals.

Five-minute integration

One curl per engine. Every endpoint accepts country_context at runtime — no hardcoding, ever.

bashEngine 01 · POST /engine/skills/extract

curl -X POST https://unmapped-614242090378.europe-west1.run.app/engine/skills/extract \
  -H "Content-Type: application/json" \
  -d '{
    "input_text": "I have been fixing phones for 5 years, taught myself Python on YouTube, manage 3 people at the shop.",
    "education_level": "secondary",
    "country_context": "ghana",
    "taxonomy": "esco",
    "language": "en"
  }'

bashEngine 02 · POST /engine/risk/assess

curl -X POST https://unmapped-614242090378.europe-west1.run.app/engine/risk/assess \
  -H "Content-Type: application/json" \
  -d '{
    "profile_id": "00000000-0000-0000-0000-000000000001",
    "country_context": "ghana",
    "automation_dataset": "frey_osborne",
    "include_landscape_projection": true,
    "projection_horizon_year": 2035
  }'

bashEngine 03 · POST /engine/opportunities/match

curl -X POST https://unmapped-614242090378.europe-west1.run.app/engine/opportunities/match \
  -H "Content-Type: application/json" \
  -d '{
    "profile_id": "00000000-0000-0000-0000-000000000001",
    "assessment_id": "<assessment_id from /engine/risk/assess>",
    "country_context": "ghana",
    "opportunity_types": ["formal_employment","self_employment","gig","training_pathway"],
    "max_results": 8,
    "include_econometric_signals": true
  }'

All three engines are independently useful. You can call Engine 03 with any external profile_id you already store; pass assessment_id only if you want automation-risk weighting in the score.

Who calls what

Pick the smallest surface area you actually need.

Who	Why	Endpoint
Employer / HR platform	Verify a candidate’s informal claims against ESCO/ISCO/O*NET in one call.	POST /engine/skills/extract + POST /engine/skills/validate
Training provider / NGO	Know what to teach next: get adjacent resilient skills + Wittgenstein 2035 landscape.	POST /engine/risk/assess + GET /engine/risk/landscape
Government / policymaker tool	Embed regional skill-desert scores and growth-leader sectors in your own dashboard.	GET /engine/opportunities/dashboard
Jobs platform	Power skill-to-opportunity matching with real wage + automation-risk weighting.	POST /engine/opportunities/match

Add your country in four steps

Zero country logic in Python or TypeScript. The full Kenya rollout below was four files. The frontend reads the registry from /engine/meta/countries on startup.

bashRecipe

# 1. Drop a new country config (one file, no code changes)
backend/config/myanmar.json

# 2. Add a row to each shared dataset (uses ISO-3 country code as key)
backend/data/wittgenstein_2035.json   # education projections
backend/data/ilo_wages.json           # wage medians + by-education
backend/data/policymaker_regions.json # regional skill desert seed

# 3. (Optional) restart backend; load_country_config() is lru_cached
uvicorn backend.main:app --reload

# 4. The /engine/meta/countries endpoint now lists Myanmar.
#    The frontend CountryToggle hydrates a new button.
#    All three engines accept country_context: "myanmar" immediately.

jsonbackend/config/kenya.json — abridged

{
  "country_name": "Kenya",
  "country_code": "KEN",
  "currency": "KES",
  "currency_symbol": "KSh",
  "language": "en",
  "greeting": "Karibu",
  "automation_lmic_calibration": 0.74,
  "education_taxonomy": {
    "primary": "Primary school (KCPE)",
    "secondary": "Secondary school (KCSE)",
    "vocational": "TVET / Polytechnic",
    "tertiary": "University degree"
  },
  "top_growth_sectors": [
    {"sector": "digital_services",  "yoy_growth_pct": 38},
    {"sector": "agritech",          "yoy_growth_pct": 24},
    {"sector": "mobile_finance",    "yoy_growth_pct": 31}
  ],
  "demo_persona": {
    "name": "Wanjiku",
    "story": "Running a phone-accessories shop in Eastlands, taught myself JavaScript on shared WiFi..."
  }
}

See backend/config/__init__.py → available_countries() for the loader. Every dataset under backend/data/ is keyed by ISO-3 country code; missing keys fail loudly so you know exactly what to seed.

Embed the policymaker dashboard

One GET, JSON in, render anywhere. Use it in a partner platform, a ministry briefing tool, or a static report.

tsGET /engine/opportunities/dashboard

// Server-side fetch from any partner dashboard
const dashboard = await fetch(
  'https://unmapped-614242090378.europe-west1.run.app/engine/opportunities/dashboard?country=ghana&region=all&view=policymaker'
).then(r => r.json())

// dashboard.regions[] => [{ region_name, lat, lng, skill_desert_score,
//   youth_neet_pct, top_missing_skills, top_unfilled_roles, ... }]
// dashboard.national_summary => { youth_unemployment_pct, neet_pct,
//   informal_employment_share, fastest_growing_sectors, data_sources }

// Drop into your own map / chart layer. Every value carries a data_source.

Or drop the policymaker page directly into an iframe:

htmliframe — quickest possible embed

<iframe
  src="https://your-unmapped-host/policymaker"
  width="100%"
  height="800"
  style="border:0; background:#080808"
  loading="lazy"
  title="UNMAPPED Policymaker Dashboard"
></iframe>

Discoverability

Self-describing endpoints. Point any client at /engine/meta/countries to find what country contexts this UNMAPPED instance supports.

bashCountry registry

# Discover which countries this UNMAPPED instance supports
curl https://unmapped-614242090378.europe-west1.run.app/engine/meta/countries

# Inspect a single country's full config (education taxonomy, opportunity
# types, demo persona, currency, greeting)
curl https://unmapped-614242090378.europe-west1.run.app/engine/meta/country/ghana

OPEN SWAGGER /docs ↗OPENAPI.JSON ↗SEE THE ENGINES IN OUR UI →