An enterprise translation memory platform in the cloud

The problem

Translation memory is the asset a localisation business is actually built on: millions of previously translated segments that have to be searchable, versioned and correct. Enterprise customers needed to manage those translation units at scale, without the tooling becoming slower as their memory grew.

How it was built

• Built the platform in React with TypeScript on the frontend and Node.js with Express behind it, so one type model described the data on both sides of the wire.
• Treated security as part of the data model rather than a layer over it, given that customer translation memory is commercially sensitive material.
• Covered behaviour with Jest at the unit level and Cypress end to end, so regressions surfaced in CI rather than in a customer's account.
• Worked directly with QA testers, backend engineers and neighbouring teams to keep integration points stable across releases.

Result

• A production platform handling enterprise-scale translation-unit management.
• Integration across several teams held stable through continuous release.
• Test coverage at both the unit and end-to-end level, maintained rather than written once.

Services involved

• React & Next.js Development

  React frontends that stay fast as the product grows.

• Node.js Backend & API Development

  APIs that hold their shape under load.

• Web Performance Optimisation

  Measured before, measured after, or it did not happen.