Case study
Relevant Search in 14 Languages for a Global Learning Platform
How SquareShift built 20 custom plugins and a 30-attribute scoring model to tune OpenSearch relevance across 14 languages for a global learning platform.
Book a session →
Client
A global provider of AI-powered learning software, used by large businesses and government organizations.
Industry
Education
Services
Stack
OpenSearch · Search Relevancy Tuning · Multilingual Content Processing
How SquareShift delivered it.
The challenge
Search returned non-contextual results across the platform's 14 supported languages, frustrating users trying to find content. Older content had no translations, so it couldn't compete with newer content in search results, and French-language compliance in Canada added another requirement search had to meet.
What we delivered
SquareShift built 20 custom plugins and a 30-attribute scoring model to tune relevance for each language individually, treating every supported language as a first-class citizen rather than an afterthought bolted onto English.
Older content was updated with translations so it could compete equally with new content in search results, instead of falling behind by default.
The outcome
The platform now returns relevant results in milliseconds across all 300GB of indexed content, in any of the 14 languages, from a single OpenSearch stack instead of a patchwork of fixes. For relevance tuning like this in your own search, see our AI-powered search.
Multilingual search fails quietly — nobody files a ticket, they just stop searching. Tuning relevance language by language is what makes search trustworthy in all fourteen, not just the one it was built for first.
Platform & Software Engineering Practice Lead, SquareShift
