Final project
Syllabus reminder. The final project is either (1) analysis of a social media dataset or (2) an analytics-based solution for a novel application. Deliverables include a written report, a final presentation, and, if you built an application, code in a GitHub repository (with a clear README so someone can find and run your project). Teams: up to six members; topic chosen by mid-semester. Strong work is creative and original and leaves a lasting impression.
Report length and depth
- Length: about 10–15 pages, single-spaced in the PDF you submit.
- Depth: clearly state the problem or application; provide rigorous, in-depth analysis of data or a detailed examination of your proposed solution (or both, if relevant).
Suggested report sections
The following is a suggested way to organize the report. Adapt headings to match your project as long as the story stays clear.
Before the main body, include a title page (title, team or author, course). After the title page, use an executive summary or abstract as described in the table below. After the final main section, add references. Use appendices for extra material that would clutter the main text.
| Suggested section | What to include |
|---|---|
| Executive summary | A short opening summary (often one paragraph) right after the title page: what you did, the main takeaway, and who the work is for. If your project includes code, include the URL of your GitHub repository here so readers see it immediately; you will repeat repo details later in Implementation, but the executive summary is where graders and readers look first. |
| 1. Introduction | Why the topic matters; problem and who it serves; market or context (audience, trends). Plain-language what the project does (inputs, outputs, user journey). Your generative AI angle: models, modalities, agents — why gen AI fits. Relevant limitations of the status quo (cost, trust, moderation, APIs). Ethics, risk, and governance (disclosure, ToS, misuse)—at least a brief treatment here or in Results and Analysis. |
| 2. Implementation | System description: major components of the app or analysis pipeline; diagrams welcome. Stack and integration: framework (e.g. Python, web), AI models and APIs, how configuration and secrets are handled (never put API keys in the PDF). Agentic flow for projects that use agents, tools, or multi-step LLM flows—diagram plus short narrative; if you did not use that pattern, say so in a sentence. Repository: link to GitHub (or equivalent), how to run and reproduce a demo. |
| 3. Results and Analysis | Evidence: screenshots, sample outputs, or (for dataset work) key figures — caption each. What works and what does not: honest critique of brittle or “AI-sloppy” behavior. Informal testing when you gathered user feedback. Economics: rough operating costs (API, hosting); pricing idea if you would sell the service. Competitive landscape: a small set of comps (similar tools or papers) and how you differ. Deeper analysis of results belongs here, not only in Conclusion. |
| 4. Conclusion | Summarize main findings and lessons. Extensions and roadmap: next steps; how better AI in the next 12–24 months could change the product; what you would need (data, evaluation, safety). Tie back to the Introduction’s problem. End with a dedicated References section (before any appendices). |
Dataset-focused projects
If you analyze a social media dataset rather than ship an app, you can follow the same structure: executive summary plus the four numbered sections. Put data pipeline and methods in Implementation; put figures (distributions, networks, topic models, gen-AI-assisted labeling workflows, etc.) and interpretation in Results and Analysis instead of app screenshots. If you have analysis code or notebooks in GitHub, still put that repo URL in the executive summary.
Final presentation
Your slides should tell the same story in about 10–15 minutes: problem → demo or key result → how AI is used → one honest limitation → what’s next. Do not paste the full report; one strong demo slide beats ten bullet walls.
Formatting tips
- Include the GitHub (or equivalent) repo URL in the executive summary or abstract so it is visible on the first page of the narrative.
- Number figures and tables and refer to them in the text (for example, “Figure 3 shows …”).
- Cite sources and tools in a consistent reference list at the end.