All selected work
- Project
- Sep 2026 - Present
- In progress
Thai Public Data Platform
Local-first data platform
Turns two heterogeneous government Excel sources into reproducible, quality-gated analytical data.
Key numbers
- government Excel sources
- 2
- PostgreSQL layers
- 4
- Airflow DAG tasks
- 8
Highlights
- ParsingSource-specific parsers handle merged cells, multi-row headers, totals, and subtotals.
- ModelingPostgreSQL raw, staging, core, and ops layers with constraints, natural-grain uniqueness, and cell-level source lineage.
- Quality gatesFail-closed checks for keys, numeric bounds, duplicate grain, source identity, row collapse, and detail-to-total reconciliation.
- ReleasesSHA-256 release identity, retry-safe publishing, and explicit replay and backfill behavior.
- OrchestrationAn eight-task Airflow DAG, with ClickHouse for analytical serving.
- EngineeringDocker Compose, CI, tests, migrations, runbooks, and architecture decision records.
A local-first project using public data. It does not claim cloud deployment or production-scale operation.
