The person behind the pipeline.

I build software for electricity markets, energy trading, and cloud infrastructure, and lately AI agents and AI-literacy tools. I lead digital and technology work in the Philippine energy sector, where I have spent years turning market data from the wholesale spot market, trading operations, and generation-company workflows into working software.

Day to day that means market data pipelines, P&L and trading-ops tooling, and dashboards for the Philippine Wholesale Electricity Spot Market; LLM-powered analytics and multi-agent orchestration; and the cloud, data, and web stack for a group of energy companies, end to end, on Google Cloud (BigQuery, Cloud Run, Cloud SQL, Vertex AI) with Cloudflare, Docker, and Terraform. I write in Python, SQL, R, and TypeScript.

This site is where I answer small questions with real data and write up what the numbers actually say, and where I keep the longer pieces about things I build.

How this site is built

By design, the same way the projects are: no framework, no black box. Static HTML on a mono design system — paper, charcoal, Inter, hairline rules — with zero build step required to read it.

• Charts come from the lieflat-en skill: a template-driven library of 64 chart types where every figure states its units, its source, and the template it was born from.
• Projects follow one scaffold: question → data contract → notebook → audited charts → essay. Each project folder is self-contained and reproducible.
• Essays only claim what a notebook can reproduce. If a number appears in the text, it came from a file.

Colophon

Type is Inter (SIL OFL). The design system is the mono grammar from lieflat-charts — a paper ground, charcoal ink, one gray ladder, no hue on the site itself (color belongs to the charts). Layout measures: 720px for reading, 980px for wide evidence. The chart library is PolyForm Noncommercial-licensed with attribution; everything else here is mine.