Architecture before decoration
I care about boundaries, data flow, failure modes, security and the boring details that decide whether a system lives or collapses.
Software architect, senior engineer and entrepreneur. I turn messy technical reality into systems that can scale, survive change and increasingly work with autonomous AI agents.

The site is built around how I actually work: deep systems thinking, shipping, experimentation and a healthy distrust of buzzwords without working code.
I care about boundaries, data flow, failure modes, security and the boring details that decide whether a system lives or collapses.
I use agents, specs, local models, LLM tooling and automation as engineering primitives, not as a novelty layer bolted onto yesterday's process.
A system that is elegant but does not move the business is still a failed system. I design with cost, people, operations and time in the loop.
Traditional engineering is still the foundation. AI-native development extends it, it does not magically repeal physics.
ARCHDistributed systems, APIs, integration, data boundaries, scalability.
WEBTypeScript, JavaScript, React, Next.js, Angular, Node.js.
BACKPython, Go, REST, PostgreSQL, MongoDB, MariaDB.
SECAPI security, application security, threat-aware architecture.
DEVOPSDocker, CI/CD, cloud platforms, observability, automated deployment.
QATDD, Playwright, Cypress, Jest, automated regression systems.
AILLM tooling, agent workflows, local models, orchestration and evaluation.
SPECRequirements turned into executable context for humans and agents.
The interesting part is not a list of titles. It is the chain of problems that got harder, stranger and more international.
I began at the intersection of languages, localization and digital workflows. That forced me to think early about structured data, automation, interfaces and systems that behave differently across cultures.
In China, Hong Kong and across Asia I moved from implementation to leadership, business operations and IT strategy. I led teams, built systems for international distribution and learned what happens when software meets factories, sales, logistics and real money.
Running an international company made technical decisions impossible to separate from cash flow, acquisition, hiring and execution. Architecture stopped being diagrams. It became a survival mechanism.
Today I work on large-scale retail web and app systems while pushing AI-native workflows, specs and multi-agent architecture. The point is not to replace engineering. The point is to multiply the amount of engineering one good system can perform.
The emerging workflow is not 'developer asks chatbot'. It is a network of specialized agents operating inside explicit constraints.
Breaks goals into work, delegates, checks state and keeps the system moving.
Coding, testing, security, content, research and review agents operate with narrow context and strong contracts.
Humans set intent, architecture, acceptance criteria and risk boundaries. Automation handles the grind.
I work across cultures because I have spent years doing exactly that. Language is not decoration on top of a product. It changes how people understand, buy and use it.