Milestones
Project Milestones - Overview
| Milestones | Brief Description | Due Date | Grade % |
|---|---|---|---|
| MS1 | Form teams and submit a project proposal focused on the problem you want to solve, why it matters, who the stakeholders are, and what meaningful impact or success looks like. Include preliminary notes on data, an initial solution direction, and a simple application mockup/wireframe (even low-fidelity) to clarify user flow and scope. Technical choices may remain high-level at this stage and can evolve in later milestones. | 9/29 | 8 |
| MS2 | Move from proposal to implementation: set up reproducible environments (VM/virtual env + containers), build a containerized data pipeline with basic data management/versioning, and deliver an initial working app skeleton that reflects your MS1 mockup and user flow. | 10/20 | 14 |
| MS3 | Integrate your model into a running backend system. Teams should have advanced training workflows (experiment tracking, serverless training on Vertex AI), a serverless deployment (Cloud Functions or Cloud Run), production monitoring, and a working API. | 11/12 | 18 |
| MS4 | Design and develop a user-facing application: an application design document, a frontend connected to the API from MS3, Continuous Integration with GitHub Actions, and an automated test suite with ≥50% coverage. Cloud deployment and scaling are covered in MS5. | 12/1 | 18 |
| MS5 | Complete the project with Kubernetes deployment, a CI/CD pipeline with ≥70% test coverage, a Medium blog post, a 6-minute video, and a live demo at the Dec 10 showcase. | 12/11 | 26 |
Detailed instructions and submission requirements for each milestone are available on their respective pages: MS1, MS2, MS3, MS4, and MS5.
Total project points: 84 (MS1: 8, MS2: 14, MS3: 18, MS4: 18, MS5: 26)
Evaluation Format
- MS1 is submitted as a written proposal (PDF on Canvas).
- MS2, MS3, and MS4 are evaluated through a 15-minute team presentation (with slides) to your assigned project TF, held during your regular weekly TF meeting. The TF will ask questions of the team and of individual students, covering both the slides and the contents of the GitHub repository. Every team member must be prepared to answer questions on all parts of the project — code, data, infrastructure, modeling, deployment, and design decisions. “I didn’t do that part” is not an acceptable answer.
- MS5 is evaluated through the final video, blog post, GitHub repository, and the Dec 10 showcase.