Final Project
Project Requirements
Project topics must include sufficient scope and apply course knowledge to a useful end. The project must implement a complete ML system that includes training infrastructure and deployment serving, demonstrating the transition from experimental development to production operation. It must compose at least two distinct computational components that demonstrate distributed system principles (examples: training and inference pipelines, data processing and model serving, or experimentation platform and production deployment). The project must demonstrate comprehensive understanding of the entire development stack and the product lifecycle from idea to deployment to maintenance. Additional requirements and guidelines will be discussed closer to the commencement of the project.
All projects must use Python as the primary language unless approved explicitly in writing by the instructor. But projects may use additional languages for tooling and support. Projects must implement and expose some API or service to consumers. The instructor may provide additional requirements when introducing the final project assignment.
Grading and Milestones
| Item | Due | Weight |
|---|---|---|
| Topic proposal (initial and revised) | week 10 & 12 | 3% + 7% |
| Status report - Design, components, integration | week 14 | 7% |
| Technical review and demo | final | 25% |
| Project report | 20% | |
| Design and source code | 35% | |
| Video | 3% |
Deliverables
Topic proposal: describe the problem, proposed technical approach, and expected outcomes. It should communicate that your topic is adequately prepared and it should outline immediate next steps. But the proposal is merely a guidepost and reasonable deviations in method, approach, and scope are expected.
Written report: summarize the topic, provide relevant background (theoretical or applied), timeline and contributions, and document challenges and extensions. It should provide discussion sufficient that an uninformed expert can understand the models, analytic decisions, outcomes, and implementation. Teams should provide quantifiable metrics to justify engineering tradeoffs, including performance benchmarks, scalability analysis, and resource utilization.
Technical review and demo: Approximately 15 minutes (depends on class size) to describe the topic problem and solution. It should provide only what is necessary to understand the what and why and include minimal theoretical background. The instructor may provide a technical reference slide-deck template that must be completed in advance of the demo session.
Source code: submitted as a GitHub repository archive file (zip). It must include README file(s) that describe the repository structure, execution instructions, deployment configuration, and infrastructure requirements.
Video: a 4-minute video that describes the topic, your implementation, and your results. You may choose to upload this to a video sharing site such as YouTube but that is not required.