About
Waylon Sittampalam. Analytics engineer in Cary, North Carolina, working on metrics that mean the same thing everywhere they're used.
I've spent 13 years in analytics — econometrics and forecasting in London, then BI and data modelling in the US. Today I'm a developer working across the modern data stack and moving deeper into governed semantic layers with AI-assisted analytics and processes.
What I care about
Business users ask the same question and get different answers, because "revenue" is defined five ways across five dashboards. Pointing an AI at the raw tables makes it worse. My answer is to treat metric definitions as code: define each one once in a governed semantic layer, put the business logic in the right layer, and make every answer checkable against it — whether it's read by a dashboard, a notebook or an AI agent. That's where the name Defined Once comes from.
How I work
I am transforming my workflow to become AI-native: I write the spec, hand AI agents the end-to-end build — exploring data, writing models, tests and documentation — and review their output to the same standard as a teammate's. I validate generated code and can explain and defend every design decision. The checks that matter are deterministic code, not the model's opinion of itself.
What I've built
- A BI model migration nobody noticed: production models moved to a new warehouse with the same numbers and security, and fixes found in one model rolled out to all of them.
- Python tooling that checks BI semantic models as code: model numbers against source data, and model definitions against naming and query standards.
- Family Learning Tracker: an AI tutor that marks my kids' answers, finds the misconception behind a mistake and targets it the next day.
- A bean roasting copilot: tracks a coffee roast in real time and tells me what to change, based on the bean and my plan.
Background
MSc in Finance and Econometrics (Queen Mary University of London) and a BSc in Mathematics (City University).