I build and write about AI systems.
Hi there, I’m Ken. I’m currently an AI engineer building enterprise AI products in Singapore. I grew up in Shenzhen, China, and spent a few years in Shanghai for my bachelor’s before moving to Singapore for my master’s degree.
Right now, my work focus on context retrieval, agent harness, evaluation, and making AI behave reliably in production. I set up this site as a space to share my notes, thoughts, and what I'm learning along the way.
When I’m away from the keyboard, I’m probably out playing badminton.
Disclaimer: All views on this site are strictly my own and do not represent any entities.
Latest writing
Read the blogs →The Complete Ontology Guide: What, Why, and How to Build and Use One with LLM
What an ontology actually is, what goes inside it and in which format, how it relates to a knowledge graph, where it pays off, how LLMs build one now — and how it gets used to answer real questions, ground agents, and keep data queries honest.
RAG Without a Vector Database: How PageIndex Works
RAPTOR and GraphRAG added structure but kept the vector core. PageIndex deletes it — no embeddings, no chunking, no vector DB. Instead it turns a document into its own table of contents and lets an LLM reason about where to look.
GraphRAG vs RAPTOR: Advanced RAG Explained with Examples
Plain vector RAG chops everything into flat chunks and loses the big picture. GraphRAG and RAPTOR add structure back — a visual, step-by-step guide to how each one works, how they differ from normal RAG, and when to reach for them.