menu_book Documentation

How TrueSkills Works

Discover the methodology behind recognizing and validating invisible labor skills.

info
Overview

TrueSkills is an AI-powered profiling system that helps people discover and articulate the valuable skills they've developed through invisible labor — work that is often unpaid, unrecognized, and uncounted in traditional economic measures.

This includes caregiving, household management, subsistence farming, community volunteering, and informal work that keeps families and communities functioning.

rocket_launch Core Mission

check_circle Recognize that unpaid work IS real work
check_circle Translate everyday activities into professional skills
check_circle Connect people to formal employment, certification, and opportunities
check_circle Calculate the market value of invisible labor

science
Methodology

ICATUS Classification

TrueSkills uses the International Classification of Activities for Time-Use Statistics (ICATUS) developed by the United Nations to categorize human activities.

We focus on ICATUS Major Divisions 2, 3, 4, and 9:
2 - Production for own use (farming, construction)
3 - Unpaid domestic services (cooking, cleaning, household management)
4 - Unpaid caregiving (childcare, eldercare)
9 - Unpaid community/volunteer work

ESCO Crosswalk

Each ICATUS activity is mapped to the European Skills, Competences, Qualifications and Occupations (ESCO) framework to translate informal work into recognized professional skills.

ICATUS Activity	Derived Skills	ESCO Occupation
311 - Preparing meals	Menu planning, nutrition, time management	5120 - Cook
411 - Childcare	Child supervision, behavior management	5311 - Childcare Worker
211 - Growing crops	Crop cultivation, pest management	6111 - Subsistence Farmer

Skills Inference Engine

The system uses deterministic mapping (not AI guessing) to infer skills from activities. When you report 20 hours/week of meal preparation, the system looks up the exact skills defined in our crosswalk database — ensuring accuracy and consistency.

timeline
The 5 Profiling Phases

1

Welcome Phase

Introduction, trust building, GDPR consent. The AI explains the process and creates a safe space for sharing.

2

Activities Phase

Discover what you DO — not what you're called. The AI guides you through activity categories: caregiving, household, production, community work.

3

Skills Phase

Transform activities into skills. Each activity maps to specific professional skills with proficiency ratings (1-5).

4

Context Phase

Understand HOW and WHY your work is invisible. Explore remuneration, recognition, barriers, and aspirations.

5

Profile Phase

Receive your complete profile with validated skills, proto-occupation matches, market value estimate, and actionable pathways.

architecture
Technical Architecture

┌─────────────────────────────────────────────────────────────┐
│                     User Browser (UI)                        │
│   Neumorphic Interface • 5-Phase Stepper • Streaming Chat    │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                   FastAPI Backend (app.py)                   │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────────┐  │
│  │VLPDataLoader│  │SkillsInfer │  │ValidationLanguage   │  │
│  │(JSON lookup)│  │Engine      │  │Engine               │  │
│  └─────────────┘  └─────────────┘  └─────────────────────┘  │
│                              │                               │
│  ┌─────────────────────────────────────────────────────────┐│
│  │    ChromaDB Vector Store (Supplementary RAG)            ││
│  │    vlp_activities │ vlp_skills │ vlp_proto_occupations  ││
│  └─────────────────────────────────────────────────────────┘│
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│              OpenRouter API (LLM Provider)                   │
│         google/gemini-2.0-flash • Embeddings API            │
└─────────────────────────────────────────────────────────────┘

Primary: Deterministic Lookup

Skills and occupations are mapped via direct JSON lookup — ensuring 100% accuracy. The AI can only cite skills that exist in our database.

Supplementary: RAG Retrieval

ChromaDB provides additional context from proto-occupation narratives during profile generation phases.

database
Data Sources

📊 VLP_ICATUS_Crosswalk.json

Maps 21 ICATUS activity codes to derived skills, ESCO categories, and proto-occupations.

21 activity entries • ~120 unique skills

📋 VLP_Questionnaire.json

5-section questionnaire structure with activity categories, question templates, and scoring logic.

5 sections • 40+ questions

🏷️ VLP_Metadata_Schema.json

Defines work context dimensions: remuneration types, recognition levels, learning pathways, barriers.

6 remuneration types • 3 recognition levels

👤 Proto-Occupation Profiles (11 files)

Detailed markdown profiles for each proto-occupation with alternative job titles, ESCO crosswalks, and career pathways.

11 proto-occupations • Full narrative profiles

neurology
AI Knowledge Base

These are the files that power the Profiler AI — click any file to view its contents.

progress_activity Loading knowledge base files...

badge
Proto-Occupations

Proto-occupations are informal job profiles that recognize the real work people do outside formal employment. Each maps to formal ESCO occupations.

VLP-HM-001

Household Manager

Operations manager, budgeting, scheduling

VLP-CC-002

Community Caregiver

Childcare, eldercare, health support

VLP-SF-003

Subsistence Farmer

Crop cultivation, animal husbandry

VLP-DC-004

Digital Content Creator

Social media, content production

VLP-HN-005

Community Health Navigator

Health guidance, system navigation

VLP-CW-006

Informal Construction Worker

Building, repair, maintenance

VLP-FM-007

Family Financial Manager

Budgeting, financial planning

VLP-EC-008

Community Event Coordinator

Event planning, volunteer coordination

VLP-MT-009

Informal Market Trader

Sales, negotiation, inventory

VLP-GW-010

Platform Gig Worker

Delivery, rideshare, task services

VLP-RW-011

Waste Recycling Worker

Material sorting, environmental work

folder_open
Project Files

Key files in the project. Click to open in your file system.

Planning & Documentation

description

PLAN_VLP_Profiler.md

Original implementation plan with 26 tasks

description

VLP_Complete_Solution.md

Complete VLP methodology documentation

Data Files

data_object

VLP_ICATUS_Crosswalk.json

Activity to skills mapping database

data_object

VLP_Questionnaire.json

5-section questionnaire structure

Application Code

code

app.py

Main FastAPI backend (~900 lines)

settings

personas.yaml

AI persona and phase prompts

api
API Reference

GET /health

Health check with data loading stats

GET /session/new

Create a new profiling session

POST /chat

Main chat endpoint with streaming response


                        {messages, model, session_id, phase, provider}

POST /session/{id}/advance-phase

Progress to next profiling phase

GET /export/{session_id}

Export complete profile as JSON