Reproduce
Fix data, code, configuration, runtime and random state so evidence can be reconstructed.
SRAI Book 5 · Chapter 1 · PU-B05-C01
MLOps Foundations
Move from an experimental model to a reproducible, tested and governed operating system with explicit evidence, release gates, monitoring, rollback and accountable authorization.
01 / LEARNING OUTCOMES
Turn an experimental model into a reproducible and governed operating system.
- Explain the governed MLOps lifecycle and its accountable decision gates.
- Construct reproducible run records, configurations and immutable artifact identities.
- Compare baseline and candidate models with aggregate and slice-level evidence.
- Test data, model and service contracts before promotion.
- Connect monitoring signals to owners, thresholds, responses and rollback.
- Apply the control discipline across Agriculture, Education, Health and Habitat.
02 / GOVERNED MLOPS LOOP
Reproduce, evaluate, authorize and monitor.
Evaluate
Compare credible baselines, protected test evidence, slices and operational contracts.
Authorize
Require explicit gates and accountable approval before an artifact advances.
Monitor
Connect production signals to response, escalation, rollback and a new lifecycle decision.
03 / CONTROLLED LABORATORY
Technical success is evidence for a decision—not authorization.
The portable notebook builds a deterministic synthetic agricultural-risk workflow and preserves its data, configuration, model, evaluation and release evidence.
Regional slice evaluation, drift measurement and rollback simulation demonstrate why an aggregate score cannot govern a production system by itself.
VERIFIED PRODUCTION UNIT
| Lesson pages | 42 |
| Notebook cells | 45 |
| Code cells | 22 |
| Reproducibility seed | 42 |
| SHA-256 checks | 37 / 37 PASS |
| Validated environments | VS Code / Colab |
Evidence establishes what was built and tested.
Run records, hashes, metrics, slice checks and contracts make the candidate inspectable and reproducible.
Accountable people decide what may advance.
Approval, monitoring ownership, escalation and rollback determine whether evidence supports controlled use.
04 / SRAI DECISION STANDARD
No artifact advances merely because it exists.
- Data, code, configuration and runtime evidence are reproducible.
- Baseline, candidate, aggregate and slice results remain visible.
- Data, model and service contracts pass before promotion.
- Monitoring signals identify an owner, threshold and response.
- Rollback is testable and linked to a prior approved artifact.
- Technical completion never silently becomes institutional authorization.
CONTROLLED RESOURCES
Read, reproduce, practise and review.
READ
View the complete GitHub production unit ↗
Audited Chapter 1
Controlled publication edition.
PDF ↗ REPRODUCECanonical notebook
Executable and independently verified.
IPYNB ↗ PRACTISEStudent assessment
Progress toward independent application.
PDF ↗ REVIEWSolutions and marking guide
Review the explained answers, assessment criteria and marking guidance.
DOCX ↗ RUNOpen in Google Colab
Run the approved notebook and inspect the governed evidence package.
COLAB ↗ DOWNLOADComplete release package
Lesson, notebook, assessment, executive, presentation, video and validation assets.
ZIP ↓ APPLYExecutive brief
Connect analytical controls to decisions.
PDF ↗