About | also available in 简体中文
Now
I’m Chaoran Liu, an applied AI practitioner based in Singapore. I build practical AI applications and share the architectures, prototypes, and lessons behind them.
My current interests span AI applications, architecture, system design, and working prototypes. I care about the full path from a useful idea to a pilot that real users can try—and from a successful pilot to a production system a team can operate and improve.
Point of view
Reliable AI is a systems problem, not only a model problem. A strong model still needs meaningful measurements, representative data, traceable evidence, maintainable software, and feedback from the people who understand the decisions being supported.
That perspective comes from building production GenAI, OCR and document-intelligence workflows, structured extraction, RAG, private model integrations, and human-review tools. My experience includes financial enterprises, but the principles apply just as well to startups, SMEs, and teams testing their first AI pilot.
Path
My route into AI began with a B.Eng. in Chemical Engineering and an M.Sc. in Industrial Engineering, followed by a move into data science and software. I started this site in January 2015 to learn in public. The archive preserves that path through R, visualization, recommendation systems, deep learning, data engineering, and applied projects.
The subject has evolved, but the reason for writing has not: explaining a problem clearly is one of the best ways to understand it, and useful knowledge becomes more valuable when it is shared.
This site
This site is for founders, engineers, data scientists, technical product leaders, and domain specialists working through the gap between an AI idea, a useful pilot, and a production system. You will find:
- Field notes with concise observations from current work;
- Guides that make an implementation reproducible;
- Case studies covering constraints, decisions, outcomes, and lessons;
- Essays that propose a practical framework or point of view; and
- Projects with working demos, source code, or implementation notes.
New writing is organized around AI engineering, evaluation and reliability, document intelligence, and AI in practice. Older data-science material remains available in the archive as a record of the journey.