I'm an LLM Architect and Machine Learning Engineer who's been building AI systems since before ChatGPT made it cool. I led my own research and production team during the earliest days of LLMs, and I've been shipping agentic workflows, RAG systems, and production-grade inference pipelines ever since.
When I'm not optimizing inference from 27s → 1s or building knowledge graph integrations, you'll find me playing music (self-taught, multiple instruments) or diving deep into psychology research.
-
LLM Systems: RAG, Chain of Thought, agentic workflows, tool usage, self-validation
-
Inference Optimization: vLLM, exLlama, GPTQ, AWQ, quantization (1900 tok/s on 13B models)
-
Production AI: Knowledge graphs + LLMs, intent detection, high-throughput deployments
-
Research: Co-authoring papers on synthetic data generation with autoencoders
-
Languages: English (Native), Chinese (Professional)
-
Education: MEng Computer Science — University of Manchester
-
Interests: Psychology, Music (multi-instrumentalist)
-
Location: Based in Malaysia 🇲🇾 / Czech Republic 🇨🇿
| Role | Company | Notable Work |
|---|---|---|
| LLM Architect | Kiwi.com | Cut inference 27s→1s, built Promptogen framework, knowledge graph + LLM integrations |
| Gen AI Consultant | Aqua Security | Agentic CVE mitigation, Semgrep rule generation, company-wide LLM workshop |
| Research Contributor | Czech Technical University | Autoencoder synthetic data generation for cancer/metabolomics research |
| Director of AI | Legend Has It Studios | Built AI department from scratch, NLG innovations, shipped to production game |
| ML Research Intern | Microsoft | Knowledge graphs + LSTMs for NLP, fraud detection for Agricultural Bank of China |
"I have a deep interest in psychology, which reflects my ability to communicate and empathize with both people, and LLMs."