Final Project

Assignment

Demonstrate your ability to use, evaluate, and govern GenAI in an applied analytic context. Work in pairs (assigned/formed based on the Week 0 survey) - one shared workflow, one shared submission, with individual accountability built in where it matters most (see AI use log below).

Submission components

One submission per pair unless noted otherwise:

1. Reproducible Python notebook - Submit as a Jupyter notebook (.ipynb) rendered as .html that runs top to bottom without errors. Runs end-to-end with assertions, documented cleaning, and formal evaluation metrics.
2. Written summary (3–5 pages) - Submit as a PDF. Required sections:
   • Introduction (use case, question, significance)
   • Data (source, size, unit of observation, summary statistics)
   • Methods (approach justification, AI tools and settings, evaluation approach, assumptions)
   • Results (findings with effect sizes, CIs, P/R/F1, kappa, cost analysis - as applicable)
   • Governance (risk level, key rules, quantitative thresholds - a short summary; see Component 3 for the full memo)
   • Reflection (what you each learned, what you’d change)
3. Governance memo - Submit as a separate PDF from your written summary. Built from your two individual Week 9 governance memos, reconciled and revised for your shared final workflow, with a quantitative risk matrix and measurable thresholds. It’s the full memo - more detailed than the Governance section in Component 2.
4. Presentation slides - Submit as a PDF export of the slides you present in Week 10.
5. AI use log - submitted individually, one per partner. Each of you documents your own AI interactions, judgment calls, and verification. This is the one component that isn’t shared because it’s meant to reflect your personal decisions, not a joint narrative. See the AI use log guide.

Presentation

10 minutes per pair: 8 presenting + 2 Q&A. Split speaking time so both partners present part of the work. Demonstrate all four competencies (understand, use, evaluate, govern) with quantitative evidence. Q&A will include at least one question directed at each partner individually, so come ready to speak to the whole project, not just your half.

Peer evaluation

During Week 10, you’ll also score 2-3 other pairs’ presentations using a simplified rubric (workflow clarity, evaluation rigor, governance substance, presentation quality), submitted via Canvas the same day. This is a separate task, not part of your own submission. Your presentation will also be scored by peers as one input alongside grading.

Rubric

The first five criteria are scored once per pair - both partners receive the same score. AI use log is scored individually since it’s submitted separately by each partner.

Criterion	Excellent (10)	Adequate (6)	Needs revision (3)
Use case and workflow	Clear, realistic with well-designed pipeline	Identified but incomplete	Vague or no pipeline
Reproducible notebook	Runs end-to-end, assertions, formal evaluation, cleaning documented	Runs but sparse documentation	Doesn’t run or major gaps
Written summary with methods	All 6 sections, formal methods, summary stats, metrics with CIs	Most sections but methods lack rigor	Missing sections or no methods
Governance memo	Risk matrix, 3+ rules with measurable thresholds, escalation	Some elements with some thresholds	Principles only
Presentation	All four competencies, quantitative evidence, engages audience; both partners speak substantively	Most elements but uneven, or one partner dominates	Missing major components
AI use log (individual)	Comprehensive, shows evolving usage and judgment	Present but incomplete	Missing

Total: 60 points (per student - each partner is graded individually on the AI use log row, and shares the same score on the other five)