Prerequisites Basic computer literacy and a willingness to document experiments.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Supervised vs unsupervised learning Linear algebra, probability and optimisation basics Train/validation/test separation Leakage, bias and baseline metrics Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools Python NumPy pandas scikit-learn
Hands-on exercise Create a leakage-safe split and compare a simple baseline against a naive predictor. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A reproducible dataset card and baseline notebook. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria The target, metric, split strategy and known leakage risks are explicit. Evidence must be reproducible and tied to Machine Learning.
Next: Feature engineering & classical models
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Missing data and categorical encoding Scaling and feature transformations Linear models, trees and ensembles Pipelines and cross-validation Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools scikit-learn pandas Jupyter
Hands-on exercise Compare at least three model families using the same validation protocol. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A reproducible training pipeline with feature rationale. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria The chosen model beats the baseline on a justified metric without leakage. Evidence must be reproducible and tied to Machine Learning.
Next: Evaluation, error analysis & explainability
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Precision/recall, ROC/PR and calibration Cross-validation and confidence intervals Confusion analysis and slices Feature importance and explainability limits Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools scikit-learn metrics Matplotlib SHAP where appropriate
Hands-on exercise Perform slice-based error analysis and record failure modes by segment. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project An evaluation report with threshold selection and error examples. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria Model selection is supported by metrics and a documented error analysis. Evidence must be reproducible and tied to Machine Learning.
Next: Model tuning & experiment management
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Hyperparameter search Regularisation and model selection Random seeds and data versioning Experiment tracking Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools Optuna or scikit-learn search MLflow Git
Hands-on exercise Run a bounded search and compare results using a fixed held-out evaluation set. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project An experiment log with reproducible training parameters. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria A fresh run can reproduce the reported experiment within expected variance. Evidence must be reproducible and tied to Machine Learning.
Next: Advanced model design & robustness
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Class imbalance and cost-sensitive learning Feature drift and concept drift Robustness and uncertainty Fairness and privacy considerations Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools scikit-learn Evidently or equivalent Dataset profiling
Hands-on exercise Stress-test the model on shifted or noisy validation slices. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A robustness assessment with mitigation priorities. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria Limitations and failure thresholds are stated clearly. Evidence must be reproducible and tied to Machine Learning.
Next: Serving, inference & reproducibility
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Model serialization and versioning Batch vs online inference Input validation and preprocessing parity Latency and throughput measurement Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools FastAPI or batch runner MLflow model registry Docker
Hands-on exercise Serve the model behind a typed endpoint and test malformed inputs. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A versioned inference service with a reproducible environment. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria Training and inference transformations match and service behaviour is tested. Evidence must be reproducible and tied to Machine Learning.
Next: Monitoring, governance & retraining
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Prediction and data monitoring Drift alerts and retraining criteria Model cards and data lineage Rollback and human review Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools Monitoring dashboard Model registry CI pipeline
Hands-on exercise Simulate data drift and define an evidence-based retraining or rollback rule. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A monitoring and governance runbook. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria Operational triggers and model ownership are clear. Evidence must be reproducible and tied to Machine Learning.
Next: Capstone: end-to-end ML system
✓ Mark this node as complete
Prerequisites Completion of the previous stage and a working local practice environment.
Learning objectives Explain the stage-specific model and its limits Apply the listed concepts using a controlled exercise Record results, failure cases and a reproducible evidence trail
Topics to learn Business objective and success metric Data lineage and evaluation Serving and monitoring design Portfolio communication Machine Learning specialization: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Tools Python ML stack MLflow API or batch deployment
Hands-on exercise Reproduce the full pipeline from a clean environment and present an error-analysis walkthrough. Context for this route: Machine Learning: build leakage-safe baselines, feature pipelines, model selection and measurable inference.
Suggested project A versioned ML project with dataset notes, experiments, service and monitoring plan. The outcome must directly demonstrate Machine Learning, not merely list tools.
Learning resources Completion criteria Results are reproducible and limitations are communicated honestly. Evidence must be reproducible and tied to Machine Learning.
End of roadmap: Complete the final project and build your portfolio.
✓ Mark this node as complete