Teach and adapt
Use the learner pilot instrument to validate pacing and independent transfer. The project companion supports existing assessments without setting institutional rules.
Use an individual module or the sequence. Original companion practice does not depend on Canvas assignments. Preserve local grading/AI-use rules and a private accessible help route.
Direct links
Link repository Markdown at a tested release/commit, or use the routes below relative to the base URL of the served reading site.
| Module | Entry | Reading-site route |
|---|---|---|
| 1 | ML and AI foundations | /lessons/01-foundations/ |
| 2 | Python workflows | /lessons/02-python-workflows/ |
| 3 | Data processing | /lessons/03-data-processing/ |
| 4 | Exploration and visual communication | /lessons/04-exploration/ |
| 5 | Regression and reliable evaluation | /lessons/05-regression/ |
| 6 | Classification and decision trees | /lessons/06-classification/ |
| 7 | Clustering and learning paradigms | /lessons/07-clustering/ |
| 8 | Neural networks with TensorFlow | /lessons/08-neural-networks/ |
| 9 | Neural NLP for real messages | /lessons/09-neural-nlp/ |
| 10 | LLMs and evaluated generative workflows | /lessons/10-llm-workflows/ |
| 11 | Fine-tuning language models responsibly | /lessons/11-fine-tuning/ |
| 12 | Deployment and monitoring | /lessons/12-deployment/ |
| 13 | Ethics and fairness across the lifecycle | /lessons/13-ethics-fairness/ |
| 14 | Bounded agents and AI workflows | /lessons/14-agents/ |
All 14 modules have individual instructor pages. Alignment conventions distinguish domain-transfer support from direct evidence.
Adoption resources
Use the Canvas linking guide, adaptable assessment criteria, teaching plan, and alignment status. The AI collaboration guide supports verified assistance and an equivalent no-assistant route. Reusable authoring and review templates are included in the complete source download and built reading site.
Preserve independence and rigor
Public feedback supports self-study; optional depth is skippable and GenAI is not required. Discussion has a solo route. If grading, publish weights/submission rules. Retain source rights, unit, availability, split, and limitations when adapting. Restricted assessments/student work stay outside a public repository.
Access and verification
Scripts and text/tables support the same underlying technical decisions without a screenshot requirement. They do not establish notebook-interface operation. Manual accessibility testing is still required on actual learner tasks.
Core Windows/Linux execution and Linux CPU framework execution have distinct scopes. Tiny mechanics checks and observed poor model outcomes support critique, not deployment claims. Read sources, framework setup, and the module-specific notes.
Author and extend
Use the reusable templates to build consistent lessons, cases, adoption records, and evidence-based accessibility reviews.
Observe the applied teaching prototype
Use the pilot protocol and owner/independent-review handoff for Modules 5, 6 and 8, which add concept figures and a real-data sandbox. The other modules have practice and solution notebooks with authored checks; the same protocol applies to them. Outcome text is unchanged. Technical solution execution is recorded separately from human learner evidence; the human pilot remains pending.