AI agent data engineering
Most AI builders cannot do the data; most data engineers cannot do agents. This is the overlap: agents acting on real data at scale, with pipeline judgment underneath.
Attention is all you need.
Ying Zhou
Real business problems, solved end-to-end with AI.
Thirty years of data engineering, now pointed at AI agents. · Agile Strategy Inc (OPC) · since 1997
What I build
AI agents that can work against real data.
Thirty years of petabyte-scale data engineering, now pointed at agentic systems. The live edge is ECE cluster monitoring via Claude skills: an agent doing real operations on a real cluster.
data pipelines
agent actuation
shipped product
End-to-end AI products
From brief to shipped product, including live AIS maritime risk-intelligence in Vessel Intel.
Agentic web and site builds
Distinctive sites shipped fast through an AI build pipeline, with concept redesigns across art, education, and healthcare.
Fractional AI operator / advisory
Where AI belongs, where it does not, and the trust architecture around limits, audit, and rollback.
Proof in the open
Click through the work; the captions name what is live and what is a concept redesign.
Work · the trajectory
—
scroll to walk the timeline →
Breadth · 1996 →
Proof, not promises
The receipts under the headline.
Platforms & tools
Profiled under load, not just imported.
System map
Data systems & tooling.
hover a star — domain, year, and proof show in the centre · magnitude = years + glow
Writing
Stories from systems that had to hold.
Ongoing challenge
What's in flight right now.
Let's build it
Have a problem worth solving?
Hard systems, real scale, work that has to hold when people depend on it — that's the work I show up for. An AI product shipped, a data pipeline that behaves, or an operator's read on where AI fits. Let's talk.
Operator mode · since 1997
