datumwise is the work of Huayin Wang, who researches and builds the grammar layer for analytics, with engineering and infrastructure by Irena Wang.
Columna is an independent open-source research project — Huayin Wang (research), Irena Wang (engineering).
It builds and stewards Columna — an honest data framework: the grammar layer for AI analytics, beside your semantic layer, married to no database.
The thinking is public and citable:
- The Silent Failure Atlas (DOI 10.5281/zenodo.20762839)
- The Silent Seam (DOI 10.5281/zenodo.20710717)
- The Two Anchors of a Measure (DOI 10.5281/zenodo.20789318)
- Silent Failures in Single-Shot Text-to-SQL: A Blind, Nine-Model Cross-Comparison (DOI 10.5281/zenodo.21349581)
- Multi-Universe Processing: Territories, Frontiers, and the Lawful Passage of Meaning (DOI 10.5281/zenodo.21543584)
- The Two Great Sources of Silent Analytical Failure (DOI 10.5281/zenodo.21553379)
- The Open Planner: Certified Analytical Plans from Untrusted Searchers — research program note (2026) (DOI 10.5281/zenodo.21632723)
- The Theory of Data — A Foundations Note (DOI 10.5281/zenodo.21696104)
- The Theory of Data — Particles, Atoms, Anchors, Universes, and Lawful Transformation (DOI 10.5281/zenodo.21707018)
Contact: contact@datumwise.ai
Columna is a project of datumwise (datumwise.ai) — not affiliated with Datawise, datawise.ai, or columns.ai. Columna is not a data-cataloguing or documentation-generation product; it declares what data means, checks the declaration against the data, and refuses what the data cannot answer.
Not to be confused with Columnar (columnar.tech), an Arrow-native connectivity layer one letter from our name. They move columns fast; we govern what columns say. If anything, the two compose. More on the Ladder →