Audit & Compliance Scoring System

Business impact

Brings defensible, repeatable quality assurance to policy-file audits: statistically sampled, rubric-scored and trended over time, on a zero-license-cost, self-hosted stack.

Problem

Policy-file quality was checked ad hoc, with no sampling method, no scoring rubric and no audit trail, so QA results couldn't be defended or trended.

Solution

Designing a zero-license-cost, self-hosted stack on the office NAS (Docker): seeded stratified sampling, a versioned YAML scoring rubric, human-in-the-loop checklists (Jotform) feeding n8n workflows, findings stored in PostgreSQL, and Metabase dashboards for leadership reporting.

Results & highlights

• Pure, unit-tested scoring and sampling functions
• Designed the database schema and migrations (000–003), including a unified issue model for scheduled audits and reactive intake
• Built for a hard constraint: the carrier portal allows no automation, so about 30–40% of checks stay human-in-the-loop by design

Competencies demonstrated

• Quality assurance & compliance: Seeded stratified sampling and a versioned YAML scoring rubric make audit results defensible and repeatable.
• Data modeling: Database schema and migrations (000–003), including a unified issue model for scheduled audits and reactive intake.
• Solution architecture: Self-hosted Docker stack: n8n workflows, PostgreSQL, Metabase dashboards and Jotform checklists.
• Business intelligence: Leadership dashboards for trend and accountability reporting.
• Human-in-the-loop design: About 30–40% of checks stay manual by design, because the carrier portal allows no automation.

Architecture