Skip to content

Instructor adoption runbook

Use the session guide, module map, assessment examples and adoption checklist.

Before teaching

Use a fresh individual Ubuntu environment. Run setup, all labs twice, and the verification command in validation. Test a non-default database, username and schema. Do not distribute your .env or a VM image containing passwords. No live university credentials or Canvas source ZIP is required.

Suggested pacing: setup/prerequisite bridge; classification; modeling/quality; lineage; infrastructure discussion; security; monitoring; retention; AI governance; capstone workshops and defense. Follow the Spring 2026 module order: classification, modeling, lifecycle, infrastructure, security, monitoring, integrated operations and capstone synthesis. Place the added AI outcome within integrated governance and the capstone. Keep dates and workload allocation in the LMS.

For each session: 10-minute concept/prediction, 25-minute guided run, 20-minute interpretation/transfer, 10-minute pair review and questions. Times are adaptable. Diagnose shell/SQL basics early. Allow accessible text evidence instead of screenshots.

External learning supplement

Use the Alation University platform guide when assigning the Data Intelligence Project. Specify the selected activity, access route, reflection and local alternative in the LMS. Keep enrollment details and training records private. Use the optional Snowflake workshop supplement for the Module 4 cloud/on-premises comparison. Set workload and any submission requirements in the LMS. Airflow activities may be selected when useful; no external account is required for the local lab sequence.

Expected technical results

Four customers, four products, four orders, five items; completed revenue 139.95. Lab 3 must build ten models and at least thirty tests. Lab 5 requires nine access assertions. Lab 6 separates five saved live client observations from a fixed synthetic incident fixture. Lab 7 exposes a mitigation tradeoff. Labs 8–9 roll back their temporary-table experiments, leaving the course records unchanged.

Common misconceptions to challenge

  • A DAG is not an authorization boundary.
  • A masked identifier is not necessarily anonymous.
  • A passed test only establishes its specific assertion.
  • A denied query can be evidence of successful protection.
  • Client observations are not an independent server audit trail.
  • Two tables or snapshots on one server do not demonstrate multiple clusters.
  • One fairness metric does not establish ethical acceptability.

Reference solutions and private grading calibration can live in a separate private repository. A folder name does not make public content private. Use the public rubric and assess evidence limitations explicitly. Rehearse recovery without destroying student databases. Use Optional Labs 12–14 for isolated server logging, two-instance batch transfer and TLS/rotation. Review their documented execution scope before teaching; they remain optional and do not demonstrate production failover or storage encryption.

Author: Isaac K. Nti. Citation and reuse terms.