How I Built Meridex: 591K Business Records Across 54 African Countries

Day 1: The Vision

Africa's business data is fragmented. No single platform gives you clean, searchable access to companies across all 54 countries. I decided to build one.

Meridex -an African business data intelligence platform.

Day 1-2: Data Pipeline Architecture

The stack:

• Neon PostgreSQL for the database (serverless, scales automatically)
• Custom scraping pipelines for data collection
• Next.js frontend with AI-powered search

The Schema

Every record includes: company name, country, sector, city, contact info, and investment signals. I designed the schema to support both simple keyword search and AI-powered semantic queries.

Day 2-3: The Scraping Sprint

Built scrapers targeting publicly available business directories, government registries, and industry databases across the continent. Key challenges:

• Data normalization: Company names in French, Portuguese, Arabic, Swahili -all needed consistent formatting
• Deduplication: Same company listed across multiple sources with slight variations
• Rate limiting: Respectful scraping with proper delays and retry logic

Day 3: Loading & Verification

591,482 records loaded into Neon PostgreSQL. Verification queries:

• Records span all 54 African countries
• Top sectors: Agriculture, Mining, Financial Services, Telecoms, Manufacturing
• Average data completeness: 87% of fields populated

The Result

A searchable platform where investors, researchers, and businesses can find African companies by country, sector, or keyword -with AI-powered recommendations.

591K+ records. 54 countries. Built in 3 days.

What's Next

• Real-time data refresh pipelines
• Investment analytics dashboard
• API access for enterprise clients