Nestosh — Case Study
Tradeasia MDM
Full Stack Developer · Nestosh · Jul 2026 — Present
Context
A global chemical trading enterprise needed one authoritative source for its product data, served headlessly to multiple regional storefronts. Chemicals do not behave like ordinary catalogue products: the attribute set differs per substance, identity is a global CAS number rather than an internal SKU, and the same product has to exist in several languages at once.
Problem
A conventional catalogue schema forces every product into one rigid table, which cannot model attributes that vary by substance. Classic EAV models that flexibility correctly, but every read then costs a JOIN per attribute — slow exactly where the platform is supposed to be fastest: the public API feeding the storefronts.
Solution
Built a hybrid EAV architecture. Attributes are authored and validated in a normalised EAV structure, then projected into PostgreSQL JSONB on write, so an API read returns a complete product from a single row with zero JOINs. Product identity is keyed on CAS numbers, and language scoping resolves dynamically per tenant. MeiliSearch backs instant global search suggestions exposed over both REST and GraphQL. Document processing — AWS Textract, Translate and Bedrock — runs asynchronously on Redis queues supervised by Laravel Horizon. The admin side is FilamentPHP v5, extended with custom Livewire components, dynamic page builders and CMS blocks.
Impact
Read performance no longer degrades as attribute counts grow, and content teams manage every market from a single portal instead of one dataset per language. TODO: add your real numbers here — for example API response time before vs after the JSONB projection, or product count and number of languages live.
Building something similar?
I'm available for full-time and contract work.