Skip to main content
EconOS

Who built this, and why?

About the builder

EconOS is a working demonstration of economics, data engineering, and product craft — designed, built, and maintained end to end by one person.

Jean-Luc Saint-Fleur is an economist by academic training and a data and analytics professional. His work sits at the intersection of economic analysis and software: building data products end to end, from ingestion pipelines through statistical models to the interfaces people actually use.

He is also a community development practitioner, which shapes the central question EconOS asks — not just what the macroeconomic aggregates are doing, but what those forces mean for households, workers, businesses, and regions. His current interests include applied AI and decision intelligence: systems that help people reason about uncertain, consequential questions rather than simply reporting numbers at them.

EconOS is his proof of that combination in one artifact. Every layer — the reproducible Python pipelines, the econometric judgment in the research modules, the backtested forecasts, and the product design of the interface itself — was built by the same person, and all of it is open for inspection.

What EconOS demonstrates

  • Reproducible data pipelines

    The pipelines/ directory ingests authoritative sources without API keys, validates every snapshot, and commits the results — git log is the data lineage.

  • Econometric judgment

    The research modules apply established measures — labor-market tightness, affordability arithmetic, composite indices — with stated assumptions and a strict causal-language policy.

  • Validated forecasting

    The forecast center publishes only models that beat a naive baseline in rolling-origin backtests, with prediction intervals derived from actual backtest errors.

  • Product thinking

    Every page opens with the real question it answers. Features exist to serve a decision or an understanding, not to fill a dashboard.

  • Engineering craft

    Strict TypeScript, zod-validated data contracts, unit-tested calculations, and CI quality gates — a malformed snapshot fails the build, never the reader.

  • Honest uncertainty

    Stale data is labeled, revisions are acknowledged, intervals are shown, and limitations sections say plainly what the numbers cannot support.

Technical stack

  • Next.js
  • TypeScript
  • Tailwind
  • Recharts
  • Zod
  • Python
  • statsmodels
  • GitHub Actions
  • Vercel

Links

EconOS is open source under the MIT license. Issues, questions, and contributions are welcome — the data registry, the methodology documents, and the pipelines are all designed to be read, checked, and improved by others.