mirror of
https://github.com/harvard-edge/cs249r_book.git
synced 2026-07-16 23:24:55 -05:00
[GH-ISSUE #1603] [Celebration] The Vol. 1 book is lovely to read and learn !! #16315
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: github-starred/cs249r_book#16315
Reference in New Issue
Block a user
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?
Originally created by @Shashank-Tripathi-07 on GitHub (Apr 30, 2026).
Original GitHub issue: https://github.com/harvard-edge/cs249r_book/issues/1603
The book is such a great read and knowledge resource !
I've just completed the book after 2 weeks of extensive studying of the vol. 1 book and this book is simply one of the best resources for anyone to get started and learn about ML Systems.
I'd like to thank everyone who has done contributions to the book, enhancing it as a learning resource and making it open-source for everyone.
I would also like to congratulate everyone on the cs249r team who has made the project better through their contributions and @profvjreddi the boss himself who has done this behemoth task 😄.
I'll go on to build a few unconventional projects, more learnings and I'll continue with the Vol.2 of the book cause that's the part I'm very interested about.
Kudos to the team and everyone involved 🚀
@profvjreddi commented on GitHub (Apr 30, 2026):
Rocky, this genuinely made my day. Thank you 🙏
Honestly, in my head the whole point of this project is helping all of you succeed. Students, contributors, readers. That's the thing. Watching you spend two weeks deep in Vol 1 and come out the other side wanting to go build "unconventional projects," that is the win. That's exactly what I hoped this book would do for someone.
And I have to call it out. You've quietly become one of the most prolific contributors in the last couple of weeks. Fixes across tinytorch, mlsysim, staffml, the labs, the kits, the instructor materials. Real, careful, substantive work. The book and the surrounding ecosystem are better because you've been in here. Don't underestimate how much that matters.
Vol 2 is going to be even more your speed I think. Much more systems-y, more building, more of the unconventional stuff. Excited to have you along for it.
Also, fair point on discussions vs. issues. The notification gap is real. Let me think about how to fix that, or if you have ideas drop them and we'll iterate.
Onwards 🚀
@Shashank-Tripathi-07 commented on GitHub (Apr 30, 2026):
Yes prof. I think we can simplify discussions by either making the GitHub one with notifications and better grouping.
Or I think we can combine the contributors and everyone related to code in one slack/discord server. Currently, we have one discord server but it's for a wide variety of purposes. We can either segregate that or build our own whichever suits your constraints.
Safe to say, It depends on the tradeoffs 😂
@profvjreddi commented on GitHub (Apr 30, 2026):
If I have one request, @Shashank-Tripathi-07, and for anyone reading this, it is this: share one thing you learned.
Doesn't have to be everything. Just one thing. One clear insight, one mistake, one realization. It may seem small at first, but if every one of us does it, it will compound. It will turn our individual progress into something larger and a collective progress. That is how this becomes more than a repo. That is how people get better, faster, and most importantly, together. 🙏
Thanks again, it's folks like you who keep me going ... even on days like this when I am tired 💪 Let's do this!
@Shashank-Tripathi-07 commented on GitHub (Apr 30, 2026):
Sure, I think there are so many learnings but here are a few ones that I think had the highest impact.
"When something is important enough, you do it even when the odds are not in your favour"
I personally went from just knowing ML, Deep Learning and LLMs theoretically to getting the confidence and knowledge to deploy them, build them and run them at scale. That confidence is what now driving the next projects I'm building.
I'm now able to understand a lot of fields from SRE, DevOps to very niche fields like GPU Programming, Model tracking and Distributed training on large scale.
One of the biggest things I've noticed is the Dwarkesh Patel's podcast that was published yesterday on building and deploying frontier LLMs and a lot of famous senior engineers who've worked a lot in Software industry couldn't get a hold of what was being discussed but I was able to understand it well and break down things.
This kind of confidence through learning in my view is the main aim of any education, the individual should not just learn words but be able to take decisions and take humanity further, and that aim is fulfilled in the best manner possible through this project.
The risk of switching from traditional Kaggle based Data Scientist/MLE to a ML Systems engineers felt very wrong in the first week (I was scared if it works out or not but with the 2nd week it felt that's it's the right path) .
A side question, can we share the development server link with the world, that's housing the current two volume book rather than the default one that has just got the MIT Preview book ?
@profvjreddi commented on GitHub (Apr 30, 2026):
Closing this with a final thank-you, Rocky. Let's do this. 🚀