Juniper Hospitality Group, a fictional multistate US employer, must protect December payroll and 2026 year-end reporting while considering a bounded AI workflow. Build six cumulative records: a typed tool-response contract; earnings mapping and W-2 exception log; variance and equity reconciliation; quarter, notice and correction record; wage/worker/fringe memo; and employee-assistant/change specification. Worked examples, fallible supplied drafts and progressively less scaffolded tasks lead to one controller packet. Python standard-library practice includes twenty offline contract cases, forty earnings codes, two register slices and two hundred synthetic W-2 rows. No live payroll system, model account, employee data or credentials are required.
Six module Decision Labs and a strategic capstone use synthetic costs and outcomes, with distinct strategies and changed evidence. The optional AI-feedback exercise supports critique and revision, not an extra certificate gate. Completion uses lessons, all listed labs to any ending and the examination. The 15-hour workload is a planning estimate with cumulative tasks counted once. Federal sources are refreshed through the November 21, 2026 record; dated applicability and access limits are explicit, including superseding overtime guidance, the November contractor announcement and California privacy transition dates. Fictional tax rates are clearly separated from real US rules. This course provides educational practice, not tax or legal advice, vendor endorsement, professional credentials or operational approval.