github-starred/cs249r_book
2
0
Fork 0
mirror of https://github.com/harvard-edge/cs249r_book.git synced 2026-07-18 17:03:56 -05:00
Code Issues 276 Packages Projects Releases 26 Wiki Activity

[GH-ISSUE #1792] StaffML: Improvement suggestions — MCQ mode, coding questions, recent-papers section, and on-device architecture content #18067

New Issue
Open
opened 2026-05-28 21:49:40 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-05-28 21:49:40 -05:00
Owner
Copy Link

Originally created by @profvjreddi on GitHub (May 21, 2026).
Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1792

Summary

Feedback from an external user on StaffML (https://mlsysbook.ai/staffml/welcome/), currently described as "similar to LeetCode for MLSys" with concept coverage that is "crystal & clear." The suggestions below are additive enhancements to broaden the platform's usefulness for students and engineers preparing for MLSys interviews.

Suggested improvements

  • MCQ answer format — Offer a multiple-choice format alongside the current descriptive answers. Many hiring platforms screen candidates with MCQs, so an MCQ mode would let students and engineers practice in the format they will actually face.
  • Hands-on coding questions (Python / C++) — Add coding exercises. Companies frequently ask candidates to implement an optimization in real time during interviews, so live-coding practice would meaningfully increase the platform's value.
  • Recent-papers section — Add a small, curated section highlighting recent high-impact papers so users stay current with the field. Example: TurboQuant (Zandieh et al., Google Research; arXiv:2504.19874), which has drawn notable attention.
  • On-device deployment & emerging-architecture content — Add Q&A and blog posts on deploying and optimizing newer architectures on-device, e.g. Mixture-of-Experts (MoE) models. Extending beyond core concepts into applied, cutting-edge deployment would broaden the platform's reach.

Notes

The same reviewer also gave positive feedback on the main MLSys book update — the expanded device-deployment and systems-side material was called a valuable, well-documented reference that colleagues working in MLSys had been lacking.

Filed on behalf of feedback received from an external user.

Originally created by @profvjreddi on GitHub (May 21, 2026). Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1792 ## Summary Feedback from an external user on **StaffML** (https://mlsysbook.ai/staffml/welcome/), currently described as "similar to LeetCode for MLSys" with concept coverage that is "crystal & clear." The suggestions below are additive enhancements to broaden the platform's usefulness for students and engineers preparing for MLSys interviews. ## Suggested improvements - [ ] **MCQ answer format** — Offer a multiple-choice format alongside the current descriptive answers. Many hiring platforms screen candidates with MCQs, so an MCQ mode would let students and engineers practice in the format they will actually face. - [ ] **Hands-on coding questions (Python / C++)** — Add coding exercises. Companies frequently ask candidates to implement an optimization in real time during interviews, so live-coding practice would meaningfully increase the platform's value. - [ ] **Recent-papers section** — Add a small, curated section highlighting recent high-impact papers so users stay current with the field. Example: TurboQuant (Zandieh et al., Google Research; arXiv:2504.19874), which has drawn notable attention. - [ ] **On-device deployment & emerging-architecture content** — Add Q&A and blog posts on deploying and optimizing newer architectures on-device, e.g. Mixture-of-Experts (MoE) models. Extending beyond core concepts into applied, cutting-edge deployment would broaden the platform's reach. ## Notes The same reviewer also gave positive feedback on the main MLSys book update — the expanded device-deployment and systems-side material was called a valuable, well-documented reference that colleagues working in MLSys had been lacking. *Filed on behalf of feedback received from an external user.*
GiteaMirror added the area: staffmltype: improvement labels 2026-05-28 21:49:41 -05:00
GiteaMirror commented 2026-05-28 21:49:43 -05:00
Author
Owner
Copy Link

@Shashank-Tripathi-07 commented on GitHub (May 24, 2026):

Highly agreed sir, we can add a MCQ section in StaffML with multiple psychological plays (to avoid making the MCQs predictable, I've seen this same in JEE Advance 🥲) .

The in-place coding questions can be built for small questions covering technical things like build a quick neural network using Numpy, quite similar to the type of questions deep-ML has https://www.deep-ml.com/problems . Along with the calculation part of the usual questions can be moved to the same type to make the interface standardized for all questions to reduce backend complexity.

Recent papers can be pulled from Alphaxiv, paperswithcode areas and can be quoted with their official link to their papers, reducing manual intervention and work load, allowing full automation for the team.

On-device deployment and Emerging Architecture content can remain manual for you or other technical staff to write blogs or explanatory videos, that will allow the required attention needed for newer things that come up. A lot of noise and signal is there in the new frontiers and your work will be needed.

Can do all of it as you want and can be deployed at whatever deadline you choose sir 🫡

<!-- gh-comment-id:4528955690 --> @Shashank-Tripathi-07 commented on GitHub (May 24, 2026): Highly agreed sir, we can add a MCQ section in StaffML with multiple psychological plays (to avoid making the MCQs predictable, I've seen this same in JEE Advance 🥲) . The in-place coding questions can be built for small questions covering technical things like build a quick neural network using Numpy, quite similar to the type of questions deep-ML has [https://www.deep-ml.com/problems](url) . Along with the calculation part of the usual questions can be moved to the same type to make the interface standardized for all questions to reduce backend complexity. Recent papers can be pulled from Alphaxiv, paperswithcode areas and can be quoted with their official link to their papers, reducing manual intervention and work load, allowing full automation for the team. On-device deployment and Emerging Architecture content can remain manual for you or other technical staff to write blogs or explanatory videos, that will allow the required attention needed for newer things that come up. A lot of noise and signal is there in the new frontiers and your work will be needed. --- Can do all of it as you want and can be deployed at whatever deadline you choose sir 🫡
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
area: book
area: collabs
area: kits
area: labs
area: socratiq
area: staffml
area: tinytorch
area: tools
area: website
bug
dependencies
format: epub
format: pdf
javascript
link-health
link-rot
priority-high
pull-request
Mirrored from GitHub Pull Request
 staffml
type: bug
type: citation
type: code
type: errata
type: improvement
type: new
type: question
vault-sli
No Label area: staffml type: improvement
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/cs249r_book#18067
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?