Assess students' ability to apply Laplace, Gaussian, and DP‑SGD mechanisms to design and evaluate privacy-preserving data-analysis pipelines. Tasks include choosing mechanisms for interactive queries and learning tasks, computing noise scales from global/local sensitivity and (ε,δ), composing and accounting for privacy budgets (basic and advanced/RDP), implementing DP‑SGD with clipping, batching and noise multipliers, and empirically reporting utility/privacy trade-offs (accuracy, RMSE, ε/δ) with justification of design choices.