Articles

Browse my articles and working papers below.

Modelith: putting the data model back in the same file as the code that builds it

A data model that lives in a diagramming tool drifts from the warehouse within a sprint, because nothing forces the two to agree. Modelith is an ontology-anchored, git-native modeler that keeps the entity-relationship model and the dbt code in one repo, under one history. You edit an ER diagram beside your dbt YAML in VS Code, generate contract-enforced dbt through a round-trip that preserves your hand-edits, and get told exactly what drifted when the warehouse changes underneath you. Covers the protected-region emitter and three-way merge, drift classification, reverse engineering from a manifest, and why the model has to be a build artifact rather than a picture.

Aug 9, 2026

One semantic contract, three stores: routing metrics, graphs, and documents without letting an LLM write SQL

A dbt semantic layer describes tables and compiles to SQL. GraphFlow extends it to describe a knowledge graph too, compiling to Cypher, so one governed contract spans both tabular data and a graph with no second source of truth. Built for DealGraph, an investment-intelligence platform where a single question can span an aggregate, a relationship, and a paragraph of rationale. Covers the cascading router, the extract-once ingestion pattern, the metricflow_entity join contract, a compiler that mirrors MetricFlow for Cypher, and an end-to-end trace with real code.

Jul 11, 2026

Building FlowProxy: a dbt Semantic Layer that BI tools can actually reach

How I built a self-hosted semantic proxy over dbt-core and MetricFlow that speaks the PostgreSQL wire protocol, so Amazon QuickSight and LLM agents can query governed metrics with the guardrails in the YAML enforced automatically. Covers the wire-protocol emulation, the semi-additive guardrail proof, auto-cube exploration, an embedded MCP server, an air-gapped CI/CD deploy model with blue/green manifest hot-swap, and the nine bugs it took to get a live QuickSight demo on AWS working end to end.

Jul 6, 2026

Keeping AI Spend Flat While Token Usage Grows Exponentially

A practical guide to routing, caching, and semantic context for DBT/SQL/Airflow teams. Covers a local routing classifier, prefix-cache-friendly prompt construction, and treating the DBT semantic layer as a retrieval index instead of a document to paste, with instrumentation that lets Power BI track cost per token rather than absolute spend.

Jun 28, 2026

PageRank as a Popularity Contest: Structural Vulnerabilities, Collusion Dynamics, and a Byzantine Fault-Tolerant Alternative

A formal critique of PageRank as an inter-subjective popularity contest incapable of surfacing marginal truth, with a proof of collusion-susceptibility for O(log N) coalitions, analysis of the institutional epistemic loop, and a proposed BFT TruthChain architecture combining independence-weighted citation scoring, Brier-anchored content quality functions, and adversarial audit governance as a collusion-resistant retrieval alternative.

Feb 22, 2026