Independent review guide
Review a fixed commit of edition 0.7.0. This resource teaches practical ML/AI through 14 modules, with Python preparation, shared scripts/notebooks, public feedback and a real-data phishing case. Assess the learning experience and the engineering evidence together. Begin at the README, then the learner entry route.
Learner journey
- Follow the readiness route as a new programmer; assess whether Python Foundation and the ML-specific bridge supply the needed skills.
- Obtain the complete source, run Module 1 and find the saved result using only the written instructions.
- Choose a classical module and a framework module. Predict, construct, run, modify and explain the central step.
- Complete a practice task before reading public feedback. Attempt a changed task without copying the worked example.
- Inspect a failure and a numerical claim. Explain its units, denominator, baseline, split and limits.
- Follow the project companion and an advanced extension; judge whether they add appropriate depth without creating hidden core requirements.
Engineering and access
Use DEVELOPMENT.md to reproduce the workflow. Check source rights, original-versus-derived data, prediction-time availability, group/time boundaries, model selection, inference contracts and replay. Compare model claims with actual reports, including unfavorable outcomes. Assess the staged phishing case as a historical benchmark, not a live detector.
Use verification scope to distinguish tested environments from other installation recipes. Record actual browser, OS, assistive technology, task, barrier, effective alternative and retest in the accessibility log. Evaluate human task completion separately from automated structural checks.
Report actionable findings
| Field | Record |
|---|---|
| Location | File/module, tested commit and specific section or code line |
| Observation | What happened, with a reproducible step or exact evidence |
| Learner impact | Who is affected and which outcome is blocked or weakened |
| Severity | Blocking, substantial, minor or editorial, with a reason |
| Recommendation | A concrete correction and how to verify it |
Keep personal/student records and confidential reports out of public issues. Use the agreed review delivery channel. The adopting instructor owns official grading, deadlines, institutional alignment and accommodations. The review should identify improvement needs without substituting new institutional policies.