Built production AI on-call investigation agent on an internal orchestration framework (LangChain-like) with tool use (function calling), prompt chaining, memory, expanding automating incident investigation coverage from 0% to 90%+ and replacing a 30–60 min/task manual workflow.

Localization Platform:

Developing translation tools around i18n and l10n.

• Led migration of localization and translation tools for 3,000+ users to a unified, secure platform implementing multi-factor authentication (MFA) and enhancing access control for employees and external partners by 99%.

• Spearheading a multi-phase data migration to transition all localization and translation data from legacy identifiers to secure internal IDs, implementing dual-write and dual-read strategies, phased rollouts, and real-time monitoring.

• Directed multi-phase migration of 100,000+ translation assets from Dropbox to Google Drive, replacing legacy storage with modern cloud APIs and reducing file-related incidents by 95% while improving reliability for global teams and partners.

• Automated translation legal review workflows by integrating Google Sheets and async processing, reducing manual review time by 70%.

• Led multi-phase, cross-functional projects from planning to execution, coordinating engineering, product, vendor, legal, and operations teams to deliver 5 major launches on schedule.

• Mentored interns and new hires, supporting onboarding, skill development, and collaborative team growth.

• Proactively identified technical risks, tracked project impacts, and communicated mitigation plans to ensure smooth execution of complex migrations.

Ocean Planning Team:

• Designed and led development of a solution for operators to identify optimal ocean container shipment routes, managing a team of two engineers to implement shipment search and override functionality for seamless, efficient deployment across all regions.

• Managed feature requirements, prioritized tasks, and resolved critical on-call issues to maintain product reliability.

• CI Product Team: PHP(Hacklang), GraphQL, ReactJS

• Improved dev efficiency by surfacing insight and solutions for failed test/build jobs using React, GraphQL and Hack.

• Led a project that scrapes, maps, and surfaces solutions for failed jobs from internal posts and wiki.

• Implemented auto-solution reply bot to answer gathered solutions when users posts in the Q&A internal group.

• Relieved team on-call load by implementing bot that close transient tasks automatically.

• System Performance, Stability, and Security Team: JAVA, SpringBoot, AWS, Redis, MySQL, Docker, myBatis

• Improved system performance and stability by 30% by separating Redis read/write, refactoring inefficient SQLs, refactored faulty cron job center, and added Shedlock to prevent jobs retriggering.

• Reduced crucial error alerts by 90% by refactoring the error codes’ handler and implementing slack notification. - Customer Support System Team: SpringSecurity, AWS RDS, Firebase, JPA, JUnit, Mockito

• Developed customer support SaaS backend(https://www.pleased.com), which provided customers with admin pages and SDK to send and receive chat, email on their website using Firebase and SendGrid.

• Built 40+ visualized metrics, analytics, and summary tables for agent efficiency and performance statistics.

• Refactor inefficient analytic SQLs and improve performance by 5 times.

• Led a team to redesign a more efficient data pipeline and ETL system structure.

• Fixed incidents within 2 hours, investigated, and proposed postmortem reports.

• Implemented load-distribution algorithm and chat reassignment mechanism to spread support agent load.

• Reviewed security scan reports and proposed fixes for the severe vulnerability.