Warning
If you plan on referencing this resolution list for your learning journey, please refer to this blog I wrote about this list talking about how I do not necessarily plan on strictly following this timeline. However, these are still excellent resources so please make use of them as you see fit. :)
Also see this repository.
I believe that there may be a lot of resources below that may be slightly unnecessary, or maybe I am missing some resources that I need to pursue. Basically, what I want from 2025 is to be:
- really good at Python
- really good at traditional ML
- really good at ML Math
- really good at PyTorch and NumPy
- good/okay-ish at ML Systems (the engineering side)
- good/okay-ish at C++ (I don't know anything about it)
- good at reading and implementing papers
- just become a solid ML practitioner in general
I'm not sure how to exactly describe the kind of role this entails, and maybe some of these are not as important as the rest. I just want to be a really good ML practitioner -- a mixture of someone who can do ML engineering, Deep Learning (NN, architecutures, etc.), NLP, and whatever it is that I may end up liking within the sphere of ML. However, I also do not want to do things with diminishing returns (or things that won't matter as much as some other aspects) since the time I have is quite limited (full-time job does not help much). Need insights from industry leaders, experts, and basically people who know way more than me (cries)
In any case, I built the following list for now.
Important
First Half of 2025
Currently, I am okay in Python, but this is not going to be enough. I need to reach an advanced level in this language. For that, I can always add or subtract resources but currently I have:
- Practical Python ~40 hours. This should be done as soon as possible, hopefully by the end of February.
- Advanced Python Mastery -> ~50 hours. This should be done by the end of the 1st half of 2024 (end of June).
Could add some more, but this is what I have for now. Maybe this is enough, honestly.
Important
First Half of 2025
Hopefully by the end of January.
Need to throughly understand everything there. Needs a lot of revision at the moment. I also need to do the additional materials (LoRA, etc.).
Important
First Half of 2025
Hopefully by the end of February.
Important
First Half of 2025
For this, I may also need to regularly refer to the numpy documentation, but that's the exact point! Should be pretty good.
Important
First Half of 2025
Should be done with this by the end of February.
May need to add more resources. Should ask people on Twitter after I'm done with this.
However, I think I can learn this by doing things most likely (by just implementing papers, etc.). Let's see. But I will complete this PyTorch book regardless.
Important
First Half of 2025
Hopefully by the end of June.
Important
First Half of 2025
- Build a translation project through the things I've learned
- Some RAG stuff ...
Add more. I should do different kinds of them to reinforce my skills. Point to note here is that this isn't quite the ML Engineering here yet; that will be later this year.
Aiming to complete at least 4 projects by the end of June.
Important
First Half of 2025
Second Half of 2025
Only start this after you're done with the objectives till the end of February.
Aiming to read & reimplement hopefully 10 papers this year.
Important
First Half of 2025
Second Half of 2025
This is like leetcode but for ML, and there are 88 questions there as of today. Aim is to complete them all.
Aiming to solve 1-2 questions each week, starting March.
Important
Second Half of 2025
-
Get the basics down first:
-
Read the Designing ML Systems book by Chip Huyen.
-
From Ray Yoo:
- dissect how big real world companies are designing their ML systems (blog posts usually).
- Pinterest ML Blog in Medium
-
Ask yourself relevant questions regarding ML Engineering:
- Do you understand the deployment process in production?
- ... add more. Need to ask people too.
Important
Second Half of 2025
-
Build 1-2 full-stack application to understand front-end, backend, database, and deployment properly. could integrate this for any of the projects in AI, actually.
-
Do at least 1 project showcasing designing, implementing, and maintaining ML pipelines if possible. or at least have some proof that showcases my ML pipeline knowledge --> reference can be the ML System book by Chip Huyen.
-
Start writing blogs on the papers you've read, and also on the projects you do.
add some more if things come to mind.
Important
Second Half of 2025
This is something important for ML in general (DL, CUDA, etc.), and I've also always wanted to learn it anyway. This is the perfect time.
Resources? Ask people on Twitter.
Important
Reach a point where you can do basic ML stuff with C++ (writing and training a simple MLP, building an autograd engine in C, add more).
Important
Second Half of 2025
I have literally no idea how to do this, but this is worth starting to consider this year. I have to do something about this! Ask people around, search for ideas and resources, etc.
- really good at Python
- really good at traditional ML
- really good at ML Math
- really good at PyTorch and NumPy
- good/okay-ish at ML Systems (the engineering side)
- good/okay-ish at C++ (I don't know anything about it)
- good at reading and implementing papers
- just become a solid ML practitioner in general
Hopefully will get a ML job by the end of this year (aiming for within my company).
Not entirely sure if this is even needed at this point to be honest, but maybe the Georgia Tech master's if I really wanna apply.
- Paul Graham's How to do Great Work