Orgelia

Orgelia is a multilingual discovery and comparison platform for intimate products and shops. The project combines a public browsing experience with a separate data engineering pipeline for collecting, normalizing, enriching, validating, matching, and loading catalog data.

Context

Product comparison becomes difficult when the same item appears across stores with different names, languages, descriptions, variants, and identifiers. Shop discovery adds another layer: locations, country coverage, map data, and structured business information need to stay consistent as the platform expands.

My role

I designed and developed the product architecture, frontend experience, catalog pipeline, analytics, and supporting automation. I also worked on the scraper and normalization strategy used to transform store-specific source data into a reusable catalog model.

Engineering challenges

• Separate source-specific scraping from the validated catalog served by the application.
• Normalize heterogeneous product and shop data across countries and languages.
• Preserve raw source records so transformations remain reproducible.
• Match products conservatively when names and descriptions differ between stores.
• Detect schema drift, locale redirects, missing fields, and scraper regressions.
• Keep product identity separate from store-specific offers and prices.

Data pipeline

The pipeline follows a staged structure: scrape, normalize, enrich, validate, match, and load. JSONL is used as a durable intermediate format before validated records are loaded into PostgreSQL.

Related writing

• Building Orgelia: architecture for a multi-country product comparison platform
• E-commerce data pipeline design
• Product matching across stores and languages
• Reliable e-commerce web scraping
• JSONL to PostgreSQL catalog pipelines