•Overview: Built a project-based code challenge platform, ONE INTERVIEW, for employers to evaluate their candidates. Candidates’ qualification and level are rated automatically and properly by the machine-learning model we built.
•Built tools that can integrate different open source libraries with our code challenge platform and extract signals from the code in different languages (Python, Ruby, Java, JavaScript, C++, swift, etc.) to generate coverage report, linter report, etc.
•Provided standardized data (ex. coverage percentages, errors, warnings, etc.) among different languages to the machine-learning model with 20% performance improvement.
•Integrated the evaluation service into the encapsulated Docker container, which is a lightweight virtual machine, to ensure isolation and performance.
•Web Development (Ruby on Rails, AWS)
•Built the whitelist to allow candidates to take multiple challenges upon requests from employers.
•Increased the website responsiveness by 70% using the combination of media queries and better assets compression to download suitable assets based on browsing devices.
•Code Challenge Performance Data visualization (c3.js & d3.js)
•Used d3.js and c3.js to build an interactive performance breakdown dashboard for employers to easily understand each candidate’s ability among different coding skills with multiple visualized charts.
•Expanding supportive language (Swift) and project-based code challenges
•Refactored the logic of supportive languages of our platform and configured the environment of the evaluation service (backend) to support swift, as well as the online editor (frontend).
•Expanded the pool of the code challenges with the corresponding solution, and ported it into different languages. (Python, Ruby, Java, JavaScript, C++, swift, etc.)