CO grading solution




Our involvement:

Automation / Testing

Project Details

The company project was about developing an evaluation system that provides automatic decisions and insights to the internal team about work candidates, such as submissions quality or deviations from an acceptable ones, which can directly impact the key metrics set for the given position. Our client planned to migrate an old and complex legacy product to a new AWS based platform. The legacy solution had a good level of automation developed using Gitlab, Docker and Java, which was not matching the automation requirements that was already applied for the new platform.

Business needs

The new realities of how remote teams work, required adequate company strategy and restucturing of the already existing processes. That required the old automation framework and pipelines to be migrated and refactored as well. Holding global online hackathons required scalable and cost effecive pipelines.

Our Solution

We had to build both Automation and Manual QA Pipelines. The solution stack included AWS Batch (serverless architecture) and Docker, Selenium running with XVFB. Shell scripts scanned submissions for predefined rule sets. Suppot was added for JAVA, Ruby and .Net Core. The costs per each grading were decreased from $50 to $0.25 due to human efforts removal in the first stages of the checking process.