Program P07 · 320 hours
Data and Model Specialization
Build and evaluate small predictive/neural models, assess adaptation, and demonstrate a controlled model lifecycle.
- Who it is for
- Learners pursuing data/model work and prepared for additional mathematics and programming
- Total effort
- 320 hours, planned at about ten hours a week
- Final project
- A reproducible model study, scoring workflow, monitoring plan, and independent defense
What is available now
Curriculum available; lessons in preparation. All 82 units are outlines: planned outcomes, planned time and selected reading. No lesson text, quiz or task is published for this program yet.
Modules in order
- A10 AI literacy and responsible judgment
- A20 Practical prompting and verification
- D10 Python and computational problem solving
- D20 Data literacy, SQL, and knowledge lifecycle
- D30 Software engineering and reproducible delivery
- MA10 Mathematics and statistics for model builders
- ML10 Predictive machine learning and responsible validation
- E10 Evaluation, experiments, and observability
- DL10 Neural learning and multimodal foundations
- AD10 Adaptation, compression, and advanced-model judgment
- B10 Backend APIs and integrated services
- S10 AI security, privacy, and governance engineering
- O10 Deployment, reliability, and operational economics
- MD10 Model lifecycle and MLOps
- CA20 Data and model capstone and research communication
Audience and entry
Learners pursuing data/model work and prepared for additional mathematics and programming. Anyone may choose this route if readiness requirements are met. No placement by age, gender, or occupation. Optional digital bridge: ten additional hours if required. Module prerequisites must be met through study or observed challenge evidence.
Outcome and scope
Build and evaluate small predictive/neural models, assess adaptation, and demonstrate a controlled model lifecycle.
Total planned effort: 320 hours. Schedule: 32 weeks at ten hours weekly. This includes reading, practice, feedback, project, and assessment. Longer calendars at lower weekly intensity are allowed without dropping evidence requirements.
Ordered module mix and schedule
| Week/session | Module | Focus | Hours | Evidence |
|---|---|---|---|---|
| 1 | A10 | AI literacy and responsible judgment | 10 | An AI opportunity and risk journal |
| 2 | A20 | Practical prompting and verification | 10 | A tested collection of reusable task instructions |
| 3–5 | D10 | Python and computational problem solving | 30 | A tested synthetic-data processing application |
| 6–7 | D20 | Data literacy, SQL, and knowledge lifecycle | 20 | A versioned dataset and searchable store |
| 8–9 | D30 | Software engineering and reproducible delivery | 20 | A reviewed application repository |
| 10–12 | MA10 | Mathematics and statistics for model builders | 30 | A notebook explaining model behavior with calculations |
| 13–15 | ML10 | Predictive machine learning and responsible validation | 30 | A reproducible tabular prediction study |
| 16–17 | E10 | Evaluation, experiments, and observability | 20 | A versioned evaluation harness |
| 18–20 | DL10 | Neural learning and multimodal foundations | 30 | A small neural or vision experiment with baseline comparison |
| 21–22 | AD10 | Adaptation, compression, and advanced-model judgment | 20 | A bounded adaptation comparison |
| 23–24 | B10 | Backend APIs and integrated services | 20 | An integrated request-processing service |
| 25 | S10 | AI security, privacy, and governance engineering | 10 | A threat model and control-evidence pack |
| 26–27 | O10 | Deployment, reliability, and operational economics | 20 | A staged pilot and operating runbook |
| 28–29 | MD10 | Model lifecycle and MLOps | 20 | A small model lifecycle pipeline |
| 30–32 | CA20 | Data and model capstone and research communication | 30 | A defended model study and handover pack |
Weekly delivery
Use two hours of explanation/demonstration, four guided practice, two independent application, one assessment/reflection, and one clinic. Adjust the mix to the module while keeping the ten-hour total. A four-hour weekend workshop plus six distributed hours is one delivery option.
Practical finish
A reproducible model study, scoring workflow, monitoring plan, and independent defense. Learner explains input choices, output checks, failure handling, and what remains outside the assessed scope. Provide a new example to test transfer rather than asking for memorized wording.
Completion and remediation
Use the assessment policy in the master framework. Record individual evidence even when practice is collaborative. Provide feedback, a targeted retry, and a new case. Do not grant a pass solely for attendance or polished AI-generated output.
Credit and onward routes
Each completed canonical module is counted once. Assess prior learning before granting exemption. P00 is orientation and does not automatically replace A10/A20. P03 graduates need forty new hours for P04 or forty for P05. Other transitions use the difference between module sets and prerequisites, not a generic discount. Finishing a route does not require continuing into a technical program.
Scenario and resource choices
Offer family/community, learning, workplace, or technical examples appropriate to the skill. Use synthetic data and no-cost practice alternatives where feasible. Disclose optional API/subscription costs before a lab. No public publishing, real customer contact, payment, or other consequential action is required for assessment.
Choosing a program is your decision; there is no placement by age, gender or occupation, and reading a lesson is not a qualification.