Instructor session guide¶
Use alongside the module learning map and each lab's own guide. Plan approximately 10 minutes for prediction, 25 for a guided run, 20 for independent work and 10 for review. These are planning estimates; record actual times during the first pilot. Longer setup and capstone workshops need separate sessions.
Before each class, rehearse on the tested course revision and identify the expected result. Provide a text alternative to any visual demonstration. Keep solution and calibration notes in an access-controlled private location.
| Session | Prepare / demonstrate | Ask / independent checkpoint | Misconception and recovery |
|---|---|---|---|
| Foundations / Lab 1 | Fresh setup; locate root; run check | Explain OS user versus database role; complete bridge self-check | Connection is not universal authorization. Resolve setup failures before advancing. |
| Lab 2 | Inspect register; show one existing classification | Student inserts a different field with owner, retention and use rationale | Classification is contextual. Inspect the inserted row and preserve student additions on rerun. |
| Lab 3 | Trace order to items to mart; inspect one test | Student explains 139.95 total and authors a test | Join grain can duplicate totals. Inspect failed test rows; do not delete the test to pass. |
| Lab 4 | Trace a model dependency | Student traces downstream copies and proposes retirement actions | Lineage is not enforcement. Use a text dependency path if the visualization is unavailable. |
| Infrastructure discussion | Compare ownership and failure domains | Student justifies one deployment choice across six criteria | One-instance snapshots are not clusters. Lab 9 is optional; the comparison is required. |
| Lab 5 | Use separate authenticated readers | Predict one denial, then explain the actual result | A denial may be successful protection. Do not weaken authentication or use administrator credentials. |
| Lab 6 | Separate live observations from fixture incidents | Trace one flag and propose a missed-event scenario | Client evidence is not server audit. Verify source labels before interpreting alerts. |
| Lab 7 | Recalculate one group metric from the fixture | Defend a decision and respond to the change-review prompt | Metric parity alone is not fairness. Check sample sizes and competing harms. |
| Capstone workshop | Trace one requirement to evidence | Peer review, revise, then defend one uncertain claim | Baseline labs do not prove another deployment. Use the four evidence labels. |
Feedback and calibration¶
Before marking, independently assess two fictional or authorized anonymized examples using the public descriptors. Discuss disagreements and record criterion-specific expectations privately. Do not infer understanding from a green runner or attractive presentation. Ask students to explain one changed condition verbally or in text.
After each session, record common difficulties and time spent without student names. Use that record to improve instructions. Students may need different forms of evidence; assess the same concept rather than requiring screenshots.
Brief reasoning check-ins¶
Use these prompts within the existing lab discussion or written interpretation. They are instructor options, not additional required submissions or a new grading scheme. Allow an equivalent short written response when an oral exchange is not suitable. Plan about five minutes per learner as an initial scheduling estimate; adjust from experience. Ask for reasoning before offering a hint.
| After | Prompt using the learner's own work | Feedback cues |
|---|---|---|
| Lab 3 | Why can joining orders to order items overcount an order total? Explain using the grain of each source. | Look for repeated order-level values across item rows and an appropriate aggregation boundary. A passing test alone is not an explanation. |
| Lab 5 | What does the observed 42501 establish for this role and query, and what does it leave untested? |
Look for an observed authorization denial, the expected policy outcome, and a limit on generalizing to other operations or identities. |
| Lab 7 | Why would matching false-negative rates not, by itself, establish fairness? | Look for denominators, other errors or harms, context and affected people; one metric cannot settle the governance decision. |
Give one supported strength, one specific gap and one next action. Let the learner revise the existing explanation where course rules allow. Record recurring misconceptions privately for the next offering; do not publish learner responses or add scores outside the announced assessment scheme.
Optional workshops¶
Labs 8–11 extend the main work. Inspect their prerequisites and deliverables before assigning them. External platform exploration follows the platform guide and LMS instructions. Optional Labs 12–14 use dedicated local instances for server logging, batch transfer and TLS/rotation. Follow their setup and recovery guide. Optional Lab 15 supplies a primary-source governance case. Do not improvise these on shared university infrastructure.
For Labs 12–14, predict the negative control before execution, inspect evidence.json after PASS, and spend at least half the session on interpretation. Audit: distinguish statement start from completion. Transfer: distinguish batch copying from replication. TLS: distinguish trust, hostname and password checks. For Lab 15, ask which missing fact changes applicability before accepting a control recommendation.
Author: Isaac K. Nti.