Semantic Quay, Inc.
The dock where meaning and data are loaded and moved.
An engineering and advisory practice modernizing banking data infrastructure with AI — metadata-driven pipelines and published semantic models, built on the controlled data models banks already trust.
The name
Semantic, and a quay.
The firm took shape during an engagement with a global investment bank — work that led, among other places, to Singapore, a city defined by its quays: the waterfront edges where cargo docks, loads, and moves on.
There, a colleague was working at the frontier of semantic data modeling. The name fuses the two. A quay is infrastructure — the point where value is transferred. Semantic Quay builds the docks where meaning and data are loaded and moved.
The approach
Banks already have the hard part. The slow part is everything downstream.
The controlled business and physical data models that banks depend on are sound. What drifts and stalls is the layer on top — the pipelines, the classification, the semantic models, and the contracts that connect them. A single governed metadata spec can drive all of it, automatically, without weakening control.
Controlled foundation
Robust business and physical data models stay the source of truth. The metadata spec sits on top of them — it references the controlled models, it never replaces them.
Metadata-driven automation
One declarative spec generates the medallion pipelines, schedules, lineage, and operational views. Change the spec; the downstream surfaces regenerate. No drift across three places.
Published semantic models
The semantic layer and data products are generated from the same spec and published as machine-readable contracts — discoverable, versioned, classified end to end.
Published, not slideware
An open reference architecture.
The practice publishes a generic, illustrative reference implementation of a metadata-driven data mesh — readable, runnable, and free to adapt. It is the credibility engine: engineering you can inspect, not a deck.
Explore the referenceone governed spec →
- 01BDM / PDMcontrolled models
- 02Bronzeraw, archived
- 03Silverconformed, classified
- 04Goldcurated marts
- 05Semanticpublished models
- 06Warehouseserving
Databricks · Cube Cloud · Snowflake — illustrative, vendor-generic.
Work together
For institutions modernizing the data layer.
Semantic Quay takes on a small number of advisory and engineering engagements around metadata-driven pipelines and semantic modeling. Inquiries are welcome.
hello@semanticquay.com