FAQ
1. Do I need to attend lectures?
Attendance is required by university policy, but we do not take attendance. In practice, students who attend regularly tend to perform better.
2. Can I miss quizzes?
Yes. We use a “best 4 out of 6” policy, which means you can miss up to two quizzes without penalty. There are no make-ups.
3. Can I use late days on the final milestone (MS5)?
No. MS5 does not allow late days or extensions under any circumstances.
4. What happens if one group member has no late days left but others do?
Late days are applied per student, not per group.
For a group submission to use k late days, each member of the group must have at least k late days remaining.
If any member does not have enough late days, the group cannot use late days for that submission.
Late days cannot be transferred, pooled, or borrowed between group members.
5. Can I request deadline extensions beyond late days?
Only in serious and documented cases (e.g., medical or family emergencies). You must contact staff as early as possible, ideally before the deadline.
6. Can I use AWS instead of GCP?
Yes. Equivalent materials are available for AWS. However, we do not cover AWS in lectures and cannot guarantee support for AWS-specific issues.
7. What counts as a valid project?
Your project must be an end-to-end solution that includes:
- a meaningful modeling component (beyond simple API calls),
- infrastructure (training, deployment, etc.), and
- a usable interface.
8. Can I just build a frontend that calls an API (e.g., OpenAI)?
No. Pure API wrappers or frontend-only projects are not sufficient.
9. What does “fully evaluable” mean?
We must be able to access, run, and evaluate all major parts of your project. Projects that rely on private or inaccessible components may lose credit.
10. How large should the project be?
Your project should be comparable in scope to a small production-ready application, not just a prototype.
11. How are teams expected to work?
All team members are expected to understand and be able to explain all parts of the project.
12. Can I collaborate with classmates?
Yes, for discussion and learning. However, all submitted work must reflect your own understanding. Excessive copying is academic dishonesty.
13. What are the rules for using external code?
Using small snippets (a few lines) is fine if properly cited. Large-scale copying without attribution is not allowed.
14. Can I use AI tools (e.g., ChatGPT)?
Yes, as a support tool. However, your submission must reflect your own understanding.
15. What happens if parts of my project cannot be evaluated?
Unevaluable components may receive reduced credit or fail the milestone. Reach out early if you anticipate issues.
16. Can I share course materials publicly?
No. Sharing or distributing course materials without permission is prohibited.
17. What if I don’t meet all the prerequisites?
If you have taken an introductory ML course and written Python code, you should be fine. You are expected to learn quickly and work independently.
18. How do I request accommodations?
Contact course staff as early as possible. Late requests may not be accommodated.
19. What if I experience inclusion or accessibility issues?
Reach out to course staff confidentially. We will help connect you with appropriate resources.
20. Is there a class GitHub repository?
Yes. You will receive access during the semester.
21. Can I audit the course?
To request permission to audit AC215, email ac215harvard@gmail.com with:
- your Harvard University ID (HUID)
- a brief statement agreeing to the auditing terms
Important:
- You must not be currently enrolled in the course. If you are listed in Canvas, you must drop the course before requesting to audit.
Auditors must agree to the following:
- In-person attendance is mandatory. Auditors must confirm their presence during the first week or will be removed.
- Academic honesty standards apply. Violations will be reported to the Harvard Administrative Board.
- No future credit. Auditors may not take this course for credit in the future.
- No submissions. Auditors may not submit assignments, milestones, or participate in projects.
- Limited access to course resources. Auditors may not use Ed Discussion or attend TF support hours.