Designed and implemented a Python-based online inference engine with customisable feature engineering and algorithm operators, streamlining the deployment of complex models, such as order address similarity detection, for TikTok Shop’s risk control, offering developers enhanced ease and flexibility.

Led the development of a real-time unsupervised clustering algorithm service in Golang, enabling the detection and blocking of malicious orders on TikTok Shop in Southeast Asia, accounting for 10% of total page view interceptions (3,000 daily) and improving algorithm developers' efficiency by 95%.

Enhanced system stability for algorithm services and the model deployment platform during high-traffic campaigns, boosting SLI from 96% to 99.8% during peak times in 2 months. This was achieved by optimizing resource estimation, implementing a customized rate-limiting mechanism, and developing improved monitoring dashboards.

Developed algorithm operators for techniques such as UMAP, DBSCAN, and Linear Regression, incorporating interactive 3D visualization and data analysis tools to enable algorithm developers to efficiently analyze datasets and fine-tune models with minimal code.

Pioneered data migration from NoSQL to SQL database for over 800,000 users. Optimized speed of migration script to be faster by 10 times using batch queries and ensured data sanity using verification scripts.

Created the first release with SQL database on production, setting the configuration of SQL database and the foundation for future SQL releases.

Led a team of 7 people to create an instructor account request portal in 6 weeks, reducing 90% of manual effort for administrator to process instructor requests and notify instructors.

Created ExcelFlex SDK using Golang and Excelize library to streamline Excel data conversion for TikTok Shops' brand partners, cutting exportable Excel report development time by 50%.

Spearheaded design of account onboarding system for key brand partners using Golang, simplifying communication cost among 4 groups of stakeholders by 80% and reducing system maintenance effort by 80%.

Led development of regional report email service for key accounts in Europe. Deployed cronjobs, optimized email lists, and established a task management database for seamless and compliant reporting in European operations.

Created TypeScript questions and solutions for frontend interview related content.

Helped in localization of the website design and translation for Chinese users.

Collaborated in a 6-member cross functional scrum team to create rankings of products in global e-commerce software and to provide sellers with greater transparency of performance of their products, improving the penetration rate for sellers with sales by 3%.

Built line chart modules using React, Less and Storybook in Data Insights team’s charting library, which reduced configuration time for each chart from 2 days to 1 hour, and enhanced extensibility of 20 usages across codebase.

Implemented a Case Management System (CMS) using Ruby on Rails, React, TypeScript and PostgreSQL to enable Lions Befrienders Service Association (LBSA) to provide counselling services to 1,000 clients every year.

Communicated weekly with clients from LBSA and CareCorner Singapore and turn the complex requirements into a digital workflow for 250 staff.