Valentinetemi/afrigen-solo

AI-powered synthetic data infrastructure for African ML engineers. Generate high-fidelity African datasets grounded in WHO + World Bank statistics, governed by OpenMetadata with full lineage, schema, catalog, and data quality validation. # AfriGen - African Synthetic Data Infrastructure > AfriGen is a synthetic data infrastructure platform for African ML engineers, designed to generate, govern, and validate high-fidelity datasets grounded in real-world statistics. ## AfriGen Demo[youtu.be] ## 🌐 Live Demo afrigen-gtht.onrender.com > Note: First load may take 30–60 seconds (Render free tier spins down when inactive). > For full OpenMetadata catalog integration, run locally with your own instance. ## 🌍 The Problem African machine learning engineers face a critical data gap. There is almost no quality training data that reflects African contexts. Our health systems, demographics, languages, and realities. Models trained on Western data consistently underperform when deployed in Africa. AfriGen fixes this. ## 🧠 What AfriGen Does AfriGen transforms real-world African statistics into validated, production-ready synthetic datasets for machine learning. It uses World Health Organization (WHO) Global Health Observatory and World Bank data to ground Gemini AI generation in real statistical distributions, ensuring outputs reflect actual African health and demographic realities instead of random synthetic patterns. Each dataset is then scored for statistical fidelity, automatically registered in OpenMetadata with full lineage and schema tracking, and made discoverable through a live catalog. Before any dataset reaches a training pipeline, it passes through a data quality layer that checks completeness, detects anomalies, scans for sensitive data, and evaluates model readiness with AI-assisted recommendations. ## 🏗️ Architecture WHO API + World Bank API ↓ Ground Truth Statistics ↓ Gemini AI (generates synthetic rows matching real distributions) ↓ Fidelity Scoring Engine ↓ OpenMetadata Catalog (schema + lineage + governance) ↓ AfriGen UI (generate, validate, download) ## 🔬 How Fidelity Scoring Works Ev …