Recent Posts
DEEPFAKE ANON GTFIH
1bfbvU
No.502780
please tell me frameworks to learn for becoming AI engineer.
i assume they are:
>PyTorch
>CUDA
what else ? please give me a roadmap or atleast basic pointers. do i need to compete in kaggle ?
EHNiuZ
No.502782
>>502780(OP)
that south delhi femboy mog this piece of gobar
Yxw6NN
No.502785
>>502782
Which one?
EHNiuZ
No.502793
>>502785
Akshansh Vats
c+7IRP
No.502805
>>502780(OP)
Im still getting there but here is what I can recommend
1. Python proficiency and math is a must
2. Take 1-2 month to get good in traditional ML
3. Practice problems on tensortonic side by side
4. Now there are two paths you can take, either go to the implementation side of things, gen ai, langchain, langgraph, n8n etc. Or foundational deep learning
Foundational deep learning itself has many domains, NLP, Computer vision, reinforcement learning, predictive maintenance and analysis. You won't be able to decide just now what you want to pursue you'll get to know it when you make yourself accustomed with the work.
I personally like computer vision, after choosing a domain there is literally no limit. Keep looking at new papers being published, keep practicing pytorch and making projects. This takes years to master
c+7IRP
No.502811
>>502780(OP)
Cuda I don't know, that's for GPU programming. Pytorch will do for your purpose. Check out daniel burke for pytorch and follow his course from scratch . For more practice try implementing some small projects from GitHub yourself
c+7IRP
No.502819
Connections matter a lot too. If you're a tier 3 guy like me it's gonna be hard, very hard. Masters is a must have in this domain
1bfbvU
No.502861
>>502819
i AM a tier 3 guy yaara. will CF and Kaggle not be enough ? i don't have time to prep for GAyTE.
OZNEmW
No.502867
>>502861
You should decide what you really want to do first.there are lot of things in "AI". Even though I am not going for ML or anything similar I know some stuff. There's data science then there is ML which has deep learning which has NLP, computer vision and other things bihari mentions. Now main thing is whether you want to be an engineer who implements and builds product or you want to be in research, research path is simple, pick your niche and pursue it till PhD is over then apply for labs. Fortunately there are labs in india now like deepmind and sarvam but again there's a lot more in usa obviously.other is implementation path which I don't know of.
OZNEmW
No.502868
>>502861
Masters is a must if you want to do research.
c+7IRP
No.502872
>>502861
Kaggle competitions are GOAT. But you'll only be able to do them with a lot of mastery. And yes they do matter a lot. Even I'm not preparing for gate nothing happens in lundia I'll apply abroad of that doesn't workout then fuck
1bfbvU
No.502884
>>502867
i am interested in computer vision implementation as well.
there's a new UGC rule, that you can do B.Tech. Hons. with 180 credits(addition project is 20 credits) and then you will be eligible for Ph.D. if you have adequate NET-JRF score. is that path not viable ? i'm aiming for IIT(NIT atleast), but that's ONLY IF i have decided to pursue research. i am only in 2nd sem, so i don't know yet.
>>502868
I'm prepared to grind Kaggle.
c+7IRP
No.502900
>>502884
Yea it's a good path, don't go for nit. Only iiit Hyderabad and IITs
