Tang Ken Yi

Apple Online Store

I was part of the China Engineering Team, and had the opportunity to work on many interesting projects, mainly on AI. - Engineered an agentic framework that ensures deterministic code quality by orchestrating Claude Code within autonomous TDD feedback loops, driving the model to iteratively refine outputs until validation passes (20–30% faster development). - Developed, containerized, and deployed remote MCP (Model Context Protocol) servers on AWS using Docker and Kubernetes. - Led exploration and deployed the team's first OAuth 2.0 authentication system with SSO for MCP servers. - Presented solutions to 30+ stakeholders across multiple time zones and authored internal technical blogs to increasing adoption and usage of newly developed tools across engineering teams.

I was part of the Gen AI Platform Team, and had the opportunity to build an internal end to end Chatbot, mainly focusing on backend. - Deployed Large Language Models (LLMs) for production utilizing Docker and vLLM, reducing inference costs and improving response latency through efficient model serving - Contributed to the development of a full-stack internal ChatGPT platform, focusing on scalable backend services and seamless user experience. - Engineered key backend features in Node.js, including file upload integration with Amazon S3, persistent chat history storage, and document-based conversational querying. - Led the development of a PowerPoint generation use case leveraging LLMs to dynamically create presentation content; integrated into the internal productivity platform. - Implemented robust observability by logging metrics and application data to OpenSearch, enabling stakeholders to monitor usage and performance trends on the dashboard.

- Upgraded the static company website using HTML, CSS and JS, leading to a 40% increase in web traffic within two months. - Implemented responsive design with media queries, ensuring seamless performance across devices and improved mobile search rankings on Google.

- Led and managed a team of 7 personnel, overseeing daily operational, logistical, and administrative responsibilities. - Planned and coordinated section-level operations, demonstrating strong communication and people management skills. - Awarded the Commanding Officer’s Coin for demonstrating quick, decisive judgment and composure under pressure during critical operational duties.

Academic Achievements Dean’s List AY24/25 Research & Projects Distributed LLM Inference: Cold-Start Latency (Final-Year Project) May 2025 – Present - Modifying the vLLM repository to analyze scaling bottlenecks and implementing a design that separates worker and executor instances to keep workers warm while executors dynamically scale with demand. Hackathons - 4th Place Winner, EasyA x BCG Hackathon 2024 - Participant, NTU CCDS Techfest 2025 Activities - Group Leader, NTU CCDS Orientation Camp AY23/24 - Group Leader, NTU Hall 1 Orientation Camp AY23/24

AY 23/24 S2 Student Exchange Programme