Learn machine learning. Build practical AI.
A practical learning resource · IT7071C companion
Start with the fundamentals and work toward real implementations. Learn through clear explanations, shared Python code, guided notebooks, and evidence you can inspect.
Follow the full sequence or enter a single module. Every module includes prerequisite support, hands-on practice, worked feedback, teaching notes, and clearly marked optional depth. No Canvas enrollment, paid AI account, or GPU is required for the core routes.
Choose your starting point
Start from the beginning
New to Python or machine learning? Begin with a small decision, run your first script, and build the foundations one step at a time.
Follow a guided path
Choose practical foundations, IT operations or evaluation-focused practice, and follow the recurring IT cases through the course.
Bring a module into your course
Link a single lesson from Canvas or another LMS. Adapt the original practice and public feedback while preserving your course's assessment rules.
Follow the learning path
Build the foundations, compare classical approaches, explore neural and language models, then connect them to responsible systems. Each module has its own entry point and prerequisite help.
Stage 1
Foundations
Stage 2
Classical ML
Stage 3
Neural and language models
Stage 4
Responsible systems
Browse the complete module directory, including the environment needed for each topic.
Work with evidence from real examples
Forecast historical bike demand. Compare wine-quality classifiers. Diagnose errors in real SMS classification. Audit an income benchmark's limitations. Run actual local language-model generation and small adapter updates.
The examples preserve observed failures as learning evidence. A functioning pipeline can reveal an ineffective model; a fluent answer can still fail its output contract. Earlier introductory fixtures are explicitly labeled separately.
Explore the worked examples · Check data and model sources · Download code and notebooks
Read, run, change, explain
Predict what a bounded change should do. Run the shared implementation from a script or notebook, inspect named results, and explain what the evidence supports. GenAI can help when permitted; verify suggestions and retain an equivalent no-assistant route.
Learn with an AI assistant · Inspect model limitations
Access the learning in a useful format
Use the reading pages, selectable code, scripts, text summaries, and numerical tables. Chart descriptions and source data accompany visual evidence. Notebook and app interfaces need their own usability checks.
Read the access routes and known limitations. Manual screen-reader, keyboard, and zoom/reflow review remains pending; automated checks do not establish conformance.
What sets this course apart
Every result in the course is checked the way a careful practitioner would check it: a stated decision, a simple baseline that is allowed to win, settings chosen on development data, and a final test used once. Poor model results stay in the reports so you can learn from them. You can run every core exercise on an ordinary laptop, without a paid account or GPU.
Ask a question or share your work · Learning paths
Reuse and contribute
Original code uses MIT; original educational content uses CC BY 4.0. External sources retain their own rights. This resource complements a course and also supports independent study; institutional credit and grading remain with adopting instructors.
Architecture · Development status · Contribution guidance
Optional complete project and advanced investigations
Start with readiness and Python Foundation, then use the separate completed capstone and advanced investigations after the relevant core modules. The 14 modules remain independent.