Singapore, Singapore
Hi there! I have a strong background in building efficient, scalable, and intelligent systems to solve real-world business challenges. I’m currently exploring opportunities in Product Management, where I can combine my technical expertise with my passion for working with and helping people. Throughout my career, I’ve gained valuable client-facing and consultative experience, and I’m excited to apply that in a more solution-oriented, user-focused role. I thrive at the intersection of technology and communication — whether it’s translating complex systems into business value, working with stakeholders to understand their needs, or designing tailored solutions that deliver measurable impact. I also enjoy tackling inefficiencies, optimizing processes, and finding the best path forward. I hold a Bachelor’s degree in Computer Science with Distinction from NUS and have worked across startups, big tech, and the public sector. Notably, my tenure at TikTok Recommendation team has endowed me with a profound comprehension of the cutting-edge technologies and sophisticated algorithms that power one of the world's most renowned recommender systems on a global scale. If any of this resonates with you, or if you simply wish to engage in a casual chat, please feel free to reach out to me! I am always keen to connect with like-minded professionals and explore potential opportunities for collaboration. Project Domain Experiences: Computer Security, Machine Learning, Computer Vision, Natural Language Processing, Sentiment Analysis, Reinforcement Learning, Data Science, Business Analytics, Generative AI, Expert Systems, Robotic Process Automation, Recommender Systems
TikTok Recommendation System, Predict service owner – a core component responsible for feature engineering, model inferencing, and video scoring to recommend videos on “For You” feed • Executed code and feature integration between TikTok's recommendation systems and the newly acquired Musical.ly, cutting manpower and maintenance cost by 50% • Proposed and implemented a series of optimization projects, achieving total >35% CPU reduction, >32% MEM reduction, >236k logical cores savings and >$500,000 annual cloud expense savings globally across different DCs • Spearheaded platformization initiatives to synchronize >2600 TikTok’s features across various business domains (e-commerce, live, game etc.), achieving feature consistency and ~80% reduction in execution time for feature-related workflows • Managed the approval of >600 feature additions and removals from the recommendation system, following a rigorous review protocol to safeguard engineering and recommendation stability prior to model training • Deployed embedding vector caching in model inferencing module, resulting in ~15% less load and ~16% less bandwidth when querying the parameter server • Enhanced TikTok’s model integration with Musical.ly by establishing robust monitoring for feature coverage and distribution, significantly improving feature management efficiency • Participated in routine on-call duties, troubleshooting live site issues and executing timely disaster recovery measures, achieving an average of >99% SLA uptime at >400k QPS for TikTok’s recommendation system • Conducted A/B, stress and canary testing to ensure optimality in user engagement and system stability metrics following each update; integrated CI diff tools and proposed platform improvement strategies to streamline engineering processes • Executed targeted CPU and memory profiling to pinpoint resource bottlenecks and built Grafana dashboards for real-time stability monitoring
TikTok Content Ecosystem Recommendation - owner of content-related recommendation architectures, such as 1) Music Video – recommend trending videos with the same music 2) Music Selection - recommend trending music for videos during editing 3) Effects / Filters - recommend trending effects/filters to be applied onto videos 4) Explore Tab - recommend multiple trending videos in one scroll-able page • Revamped the Music Video recommendation system to leverage an advanced framework, minimizing architectural complexity and improving overall stability, resulting in a video accuracy rate exceeding >99.5% • Administered many aspects of the Music Video recommendation deployment Ops, encompassing compilation, containerization, resource allocation, global cloud deployment, A/B testing, CI/CD, stability monitoring, troubleshooting, disaster recovery, and documentation • Enhanced creator-side A/B testing module driven by the cutting-edge Counterfactual framework; incorporated fail-safe mechanisms and comprehensive logging features for increased reliability • Set up disaster recovery strategies, automated alarms and pre-experiment CI verification, collectively achieving SLA uptime of >99.9% • Segregated video scoring module and config files within the Explore Tab recommendation system, effectively reducing coupling and increasing degree of modular independence • Optimized an internal recommendation framework by building a debugging and filtering tool used during video ranking and video recall stage respectively, cutting down average CPU utilization by 13% and latency by 10% • Spearheaded the transformation of >70 static downstream clients into dynamic clients and formulated a comprehensive SOP template for the creation of new downstream services, improving runtime adaptability and efficiency • Developed an integrated monitoring system for downstream services of TikTok recommendation, facilitating efficient stability tracking tailored to each downstream classification
• Taught theory and practical Software Engineering concepts in weekly tutorials to CS students • Evaluated and graded the deliverables as well as module project of CS students
• Qualified as advisor in May 2019. Promoted to senior advisor in May 2020. Awarded the Best advisor award in August 2020 • Assessed and evaluated multiple software applications built by students • Provided constructive technical and non-technical feedback, and offered advice for improvement • Acted as primary communication channel between module coordinator and students
• Fin-tech startup providing regulatory workflow platform for document exchange on automated repository • Designed, built and deployed company’s website with server from scratch • Built a backend data-mining app with fully functional interface and database • Implemented web scraping of text and files from multiple URL sources • Migrated database to PostgreSQL virtualized in Linux Docker container • Wrote RESTful APIs and documentation with emphasis on future maintainability