Warehouse / reporting architecture
Data systems organize source data, models, QA checks, dashboards, and AI workflows into durable decision infrastructure.
What this system does
Creates data models, views, QA checks, and dashboard-ready outputs from product, finance, operations, and workflow data.
Separates source data, cleaned tables, fact/dimension logic, reconciliation, and business-facing views.
Supports AI workflows by giving LLMs and agents structured, trusted context instead of loose raw files.
How I use it
BudgetDB is the flagship example: Postgres models for vendor spend, software cost per employee, headcount allocation, QA checks, and executive outputs.
Chorus schema shows how product content can be modeled in Supabase/Postgres tables.
Ghost-AI architecture separates relational metadata in Postgres from generated artifacts in Vercel Blob.
Related projects
Evidence or examples
software-cost-allocation.sql creates analytics schema views for company software cost, team headcount, and allocated cost.
chorus-schema.sql defines platform categories, platforms, skill categories, skills, lessons, and content sources with RLS enabled.
Ghost-AI architecture notes store metadata in Postgres and generated artifacts in Vercel Blob.
raw sources -> cleaned dimensions/facts -> analytical views -> QA checks -> dashboards -> AI summaries.CREATE VIEW analytics.v_software_cost_allocated_by_team_2025 AS
SELECT team, team_employees_2025, software_cost_per_employee_2025,
ROUND(team_employees_2025 * software_cost_per_employee_2025, 2) AS allocated_software_cost_2025
FROM analytics.v_employee_count_by_team_2025
CROSS JOIN analytics.v_software_cost_per_employee_company_2025;Postgres stores metadata and access relationships; blob storage stores large generated files such as canvas snapshots and Markdown specs.