Zhan Yu

• Scaled global logistics planning to 50K+ requests/month by building Java/Kotlin services on AWS (Lambda, Step Functions, DynamoDB) for reservation/booking & consolidation workflows; added idempotency, retry/backoff, and SLO-based alerting for resilience. • Implemented streaming capacity management on AWS with Kinesis Data Streams, Lambda, and Step Functions to auto release/reassign capacity on lifecycle changes; propagated updates for immediate reuse and faster confirmations. • Reduced code-to-production from 14.7 → 4.5 days (3.3×) by optimizing org-wide CI/CD: integration tests 30 → 140+, automated validation/rollback, removed manual approvals; leveraged Amazon Q and Claude to speed reviews and test scaffolding.

• Implemented back-end system for Taobao and Tmall shopping platform’s VIP module by Java, MySQL and Spring, which ultimately led to a 3% increase in the number of monthly registered VIP members. • Coupon Rule Validation: Developed a VIP coupon validation tool to improve query efficiency for product, merchant, and coupon availability by automating coupon templates and configurations validation. Utilized strategy pattern and JSON Path to streamline rule validation, enhancing system scalability. • Optimized Merchant Backend System: Developed a multi-level approval system, allowing approval personnel to handle activity sign-ups by industry; Built a report module for merchants to track real-time product performance and display the revenue gained from participating in the campaign.

• Built a system to monitor and ensure timely replenishment of shelf inventory by computer vision techniques • Implemented deep learning models with TensorFlow and OpenCV for accurate product detection and classification, training CNNs to recognize product labels, improving inventory tracking reliability.

• Boosted WiFi availability by using C++ to build network detection in OpenHarmony operating system. • Increased WiFi robustness by 33% by using Javascript to generate availability report of WiFi indicators.

• Built an automatic response verification platform utilized by whole company groups to test updated services. (React to build pages to get user requests, such as script code, AppID, and show the validated results; SpringBoot to run script code to validate response and send email notification when invalid). • Providing 25% more search recommendation results for users by integrating various language features into the keyword-parsing module to optimize search functions.

• Improved ad accounts management by developing organization modules in the advertising platform by Java and Spring. • Ensured the stability and reliability of the advertising platform by implementing unit tests for advertising-operations-related modules.

• Presented millions of TikTok advertisers with their advertising results divided by countries and regions by adding new HSQL aggregation tasks in the big data platform and using ElasticSearch to interact with data. • Built big data flow tasks that deal with billions of data volumes, such as across-domain data services and data synchronization in a new compliance computer room to comply with the compliance policy of restricting data outflow in some countries by SQL, Java, and Spring. • Ensured the integrity of ads data by aggregating offline data by multi-dimensions to cover the real-time data.

• lmproved the management of more than 1000 daily enterprise Didi(Uber-like)orders by expanding backend modules. • Allowed employees to customize personal information by optimizing the information module in the Enterprise app. • Developed a reservation system to allow outsiders to reserve company visits and parking spaces by SpringBoot and React.