The Project

On this page

Teams build a machine learning application/system thatsolves a real problem of their choosing, running the full arc over the semester. Module 1 and Module 2 sessions feed into the project as it progresses through phases.

What each team does

  1. Identify a real problem worth solving with ML.
  2. Scope it — decide whether and how ML helps, identify downstream actions, and define success.
  3. Formulate the ML problem and its baselines.
  4. Build baselines and ML models
  5. Evaluate and make recommendations for decision-makers and users.

Examples of Projects

GenAI in the project

Every team:

  • Implements an LLM zero/few-shot baseline as part of the baseline step
  • May use LLMs for one or more pipeline stages (data collection, data wrangling, linkage, labeling, feature generation, modeling, etc.) — evaluated against human ground truth.
  • Writes a short GenAI-use reflection: where the tools helped, where they misled, and how the team verified their output.

Project timeline

Each milestone is a Canvas assignment — see Assignments & Grading.

Due Milestone
Thu Aug 28 Team creation
Mon Aug 31 Submit project idea
Mon Sep 14 Project proposal and scope
Tue Sep 15 3-minute project pitch
Thu Oct 1 Implement baseline(s) — including the LLM baseline
Tue Oct 6 Project update presentation
Tue Oct 20 Initial ML solution
Thu Nov 5 Evaluation
Thu Nov 19 Iteration 2 — add one component from Module 2
Tue Dec 1 Project presentation
Tue Dec 8 Project writeup and demo

Add one component from Module 2. For the final writeup, every team extends its system with one Module 2 topic applied to their problem — a fairness/equity audit, an interpretability or uncertainty analysis, a robustness/shift check, a field-evaluation design, or a foundation-model / agent component — and reports what they found.

Deliverables & rubrics

See Assignments & Grading for the full list of Canvas assignments, due dates, points, and the rubric applied to the final writeup, demo, and presentation.