Nyan Lynn Tun

Six years in, the work rhymes. Data arrives somewhere inconvenient, four teams need it in four shapes, and nobody agrees on what a customer is. The tooling is never the hard part — the modelling is, and the discipline to keep raw and modelled data from bleeding into each other.

About me

Model first, tool second. A clean dimensional model outlives the platform it was built on. I’ve migrated the same star schema across three engines; I’ve never migrated a pipeline that skipped the modelling.

Tests are part of the pipeline, not a phase after it. Contracts on every model, failures that page someone, and a bronze layer nobody is allowed to edit. Trust is the actual deliverable — a warehouse people don’t believe is just expensive storage.

Write it down while it’s still fresh. The reasoning behind a decision decays faster than the code implementing it. That’s most of why this site exists.

What this site is

Two halves that point at each other. Projects are full case studies — the problem, the model, the decisions I’d defend, and the ones I’d take back. They’re deliberately undated: I revise them as the work changes, so a date would only ever be wrong.

The blog is chronological and looser — what I’m learning, what broke, what I got wrong in public. Posts link into the projects they came out of, and every page lists what links back to it, so a thread can be followed in either direction instead of only forwards.

Working together

I take on consulting work around pipelines, lakehouse architecture, and data reliability — usually where something already exists and has stopped being trustworthy. If that sounds familiar: hello@nyanlynntun.com.