Worked on the microsoft/roosterjs repository to deliver a performance and correctness improvement for invisible Unicode handling in a front end editor. Leveraging JavaScript, React, and TypeScript, centralized the sanitization process by introducing a sanitizeInvisibleUnicode utility, moving invisible Unicode stripping to the editor initialization phase. This approach removed the need for repeated sanitization during DOM-to-model conversions, reducing overhead and improving editor startup times. The solution preserved intentional user input by stripping only raw invisible characters, while maintaining runtime XSS protections. The work enhanced maintainability and testability of the sanitization logic, resulting in a more efficient and robust editor workflow.