« CASE STUDIES

Rio Tinto: Development of a Universal Translator

Mechanical Rock was engaged to deliver a production-ready AI based translator for users in Mongolia and Argentina.

The Challenge

Rio Tinto is a leading global mining and materials company, with a global workforce operating across 6 continents in 35 countries. To support their global Workday implementation, they required a secure on-demand translation service capable of translating Workday and ServiceNow web pages.

Following a successful proof of concept, that included piloting the use of Amazon Bedrock which offers a broad set of capabilities to build generative AI applications with security, privacy, and responsible AI, Mechanical Rock was engaged to deliver a production-ready AI based translator for a pilot implementation to users in Mongolia and Argentina.

Solution and Key Features

Mechanical Rock developed a secure, performant web browser extension for Chrome and Edge using Amazon Web Services (AWS) within an aggressive timeframe including:

  • Multi-language detection and translation
  • Automated triggering of the translation service
  • Streaming of translation for high volumes of content
  • Ability to customise translation for Rio Tinto specific terms or names
  • User interface aligned to Rio Tinto’s brand guidelines
  • Automated deployment to Rio Tinto’s Standard Operating Environment (SOE)
  • Support for translation between 75 languages for other use cases

Screenshot of the Universal Translator Browser ExtensionScreenshot of the Universal Translator Browser Extension

Key Benefits

“A translation service that is accurate, fast and reliable”

Improved system performance and reliability

Mechanical Rock overcame networking constraints to improve translation speeds by 96%, reducing processing time from 90 seconds to less than 3 seconds. Additionally, 83 unit tests were implemented to guarantee full coverage of frontend business logic, ensuring stability and reducing the likelihood of future failures and future maintenance costs.

Enhanced User Experience and Efficiency

Mechanical Rock resolved proof of concept issues, eliminated redundant API calls and implemented the ability to add custom terminology to preserve business-specific terms and minimise incorrect translations.

Scalable and maintainable infrastructure

Mechanical Rock implemented CI/CD pipelines with Infrastructure as Code (IaC) and automated testing, ensuring reliable deployments. Pull Request (PR) status checks automatically detect backend changes, preventing failures.

Embedded Client Model for Success

Comprehensive developer documentation enables Rio Tinto’s team to maintain the solution independently.