Meet with me or any of the TAs to checkoff your project idea, we will mainly be assessing appropriateness of the topic and scope. The topic can be anything that involves both deep learning and engineering.
It does need to involve engineering. Ideally that is accomplished by leveraging some of the physics/engineering integration techniques we’ve discussed this semester, but it could use just traditional deep learning methods as long as your project is an engineering application and the use of traditional methods is appropriate (e.g., you have lots of data). This does not mean that there needs to be physical laws you can identify or something like that. Algorithm development is an option, implementation for specific hardware, etc. Engineering is a broad net.
You might consider projects that integrate with your research, or reproduce an interesting paper you find online. Projects that integrate with other courses are fine as long as you are not double counting. An appropriate scope is a project that is equivalent to 2-3 homeworks (i.e., should represent 2-3 weeks of equivalent work effort).