ph pulse

2025 · data pipeline

philippine socioeconomic data made queryable, pipeline to warehouse to insight.

python ingestion pipelines, dbt transformations, bigquery as the warehouse, and gemini-powered insights on top. the whole path from raw source to answered question, built as one system.

python · dbt · bigquery · gemini

problem: philippine socioeconomic data is public but scattered across agencies, formats, and release schedules, so every question starts with a week of cleaning.
pipeline: python jobs pull each source on its own schedule into bigquery, and dbt models turn the raw tables into tested, documented ones, region by region and year by year.
insight: gemini sits on top of the modelled tables and answers questions in plain language, showing the query it ran so the answer can be checked.
why build it whole: owning every step from raw file to answer meant every number could be traced back to where it came from.

updated oct 2026next