Jun 2024 - Present

• Engineered a greenfield, high-throughput Go asynchronous rule engine to improve the detection of TikTok Shop API traffic abnormalities by collaborating with a team of two engineers on core algorithm design and architecture, achieving global performance of over 520,000 msgs/s (400k in SG DC, 60k each in EU and US DCs) with sub-100ms P99 latency.
• Implemented a dynamic feature switch to decouple a live C++ synchronous engine from its downstream dependencies, enabling a phased, zero-incident migration that redirected traffic through a new Kafka MQ and the Go engine.
• Optimised query performance for cross-functional teams by creating a pre-joined, aggregated Hive table using PySpark, which eliminated complex join operations in common Hive SQL queries which resulted in a 65% reduction in Spark job execution time and significantly improved data accessibility and team productivity.
• Prototyped and built a feature store MVP for security detection, by persisting and serving actionable intelligence (e.g., user/device fingerprints) from a NoSQL database in under 50ms (P99) during initial testing, validating its design for enabling automated threat responses such as issuing CAPTCHA challenges.

Tech Stack

Jan 2024 - May 2024

• Evaluated the viability of a Flink-based rule engine for TikTok Shop by building an MVP and performing trade-off analysis, providing key data that informed the strategic decision to adopt the Go asynchronous engine.

Tech Stack

Nov 2022 - Jul 2023

• Architected, developed and deployed several new microservices, facilitating seamless bank transfers or credit/debit card payments between ByteDance and esteemed financial institutions like Citibank, Kasikornbank, AppotaPay, and Maybank.
• Integrated new endpoints with 90% unit test coverage by leveraging the capabilities of Citibank and JP Morgan backend systems, introducing enhanced functionalities for the ByteDance backend system.
• Implemented request forwarding services and facilitated the deployment of microservices onto a new cluster, playing a key role in TikTok's Project Texas to enhance the security of US data.
• Assumed on-call responsibilities for a week, diligently monitoring critical alerts and warnings across the entire backend systems of banking microservices, and successfully developing effective mitigation plans in response.

Tech Stack

May 2022 - Aug 2022

• Migrated deprecated AngularJS frontend to Angular 12 for Ascenda's global loyalty platforms, ensuring compatibility with Google's patches and maintaining security standards for major banks and airlines handling sensitive information.
• Engineered highly customizable and reusable Angular components, empowering developers to seamlessly integrate client-specific functionality and styling requirements, minimizing code duplication and enhancing scalability.
• Collaborated with the team by conducting thorough code reviews, ensuring adherence to high-quality coding standards, thereby reducing QA time and enabling the team to deploy new features rapidly.

Ascenda serves many clients worldwide. Here are some of the clients' websites powered by Ascenda.

• OCBC
• CTBC
• FAB
• AMEX x Hyundai
• JAL
• China Eastern

Tech Stack

May 2021 - Aug 2021

• Refactored the React frontend and Rails backend of the GIC Sparks & Smiles (Sparks) application, resulting in a more than 10x decrease in loading times, enabling efficient mentoring of youths from high-needs households by dedicated volunteers.
• Enhanced Sparks' continuous integration pipeline by implementing Docker layer caching, resulting in a notable 50% reduction in pipeline run-time, significantly boosting developer productivity by enabling faster CI runtimes.
• Migrated the Rails backend of the Active Ageing Centres (AAC) application to Go, seamlessly integrating production data with the enhanced Go API to facilitate a smooth transition of 10 eldercare centers onto the new backend infrastructure.
• Streamlined volunteer matching for the AAC application by developing new React pages and Go API endpoints, enabling social workers to efficiently match volunteers with lower-income seniors, thereby reducing management overhead.

We wrote detailed write-ups on our work during the summer. They can be found here.

• AAC
• Sparks

Tech Stack