Problem
Assessment teams need trend analysis, completion monitoring, skill-level views, and leadership-ready recommendations, but public portfolio work cannot expose real students, rosters, grades, LMS exports, or private institutional artifacts.
Case Study
A public-safe workflow for turning private-reference assessment patterns into synthetic data, SQL-backed extracts, validation artifacts, dashboard-ready outputs, and decision-support reporting.
System Story
Assessment teams need trend analysis, completion monitoring, skill-level views, and leadership-ready recommendations, but public portfolio work cannot expose real students, rosters, grades, LMS exports, or private institutional artifacts.
The project separates private profiling from public output: private-reference distributions inform synthetic generation, then validation and export steps produce public-safe data for a static dashboard and reporting artifacts.
The source repositories use Python and DuckDB for LMS-to-SQL extraction, star-schema marts, and validation; SQL for assessment extracts; R for synthesis and modeling; and small Python utilities for privacy validation and dry-run AI-assisted memo generation.
The hosted dashboard demonstrates course, grade, teacher, section, growth, completion, skill, and distribution views using synthetic assessment records. The source workflow now also produces SQL-backed extracts for course-section performance, growth diagnostics, non-participation analysis, roster QA, and readiness reporting.
DuckDB remains the reproducible local warehouse, while the optional Supabase extract path demonstrates a hosted public-safe serving contract. Dry-run AI review then summarizes deterministic extracts without becoming a data-generation step.
The privacy boundary is strongest when synthetic generation, validation, dashboard export, and reporting language are designed as one pipeline. That makes the public artifact useful without implying it contains real school records.
Evidence
The case study is backed by a hosted dashboard, a public R gradebook synthesis report, and validation summaries that document the public/private boundary.