First rule of fight club is:

I am not a pro in this, so if you don’t like my advice just fck off bitch. Also read the whole blog, there are many types of questions which may help you. It covers resources to jobs typically everything.

If you’re a senior or carry more experience in ml then think of me as your junior bro, correct my mistake by commenting below my X posts.

ML = machine learning HOML = Hands on ml with sklearn, keras & tensorflow HOLLM = Hands on Large language models TF = Tensor flow

That’s being said, let’s begin :

⇒ No roadmap bro, i just fck around & find out shit. See my portfolio books section, those are the books i followed. For beginners in ml, don’t be afraid to explore other domains, in tech you never know what will excite you, so yeah explore bro. Also don’t stay stuck too long, if you can’t understand something just say fck it & move forward, trust me i have done it so many times. Like in ml when i was studying SVM, when i couldn’t understand it’s derivation i simply move forward.

⇒ Depends on so many things bro, like how deeply you’ve studied, do you read papers along it, do you write code with it or not. For HOML book, it took me around (45-60) days only in Deep learning section as i have studied core ml from other book that python data science handbook. Mind you, i daily read it around 6-7 hours & coded along the book. So yeah it will take some time, as that book & concepts are so deep. I think you need to at least finish HOML, HOLLM, AI engineering, LLM engineer handbook & some connections to get an internship. Though there is no fixed path in tech. Also for research, you need math as well.

⇒ Bro i don’t know about bootcamp, but i think you should read homl for sure, that book is legit so good, also you’ll develop a habit of book reading which will help you in further book reading like HOLLM, & others. Also HOML contains link to research papers along the topics as well as in the end of chapter, but i recommend you to read book only after this book, you can read papers as well.

Another beginner tip is that go in order like python, numpy, pandas, matplotlib, scikit learn, then HOML. Go through my portfolio books section, from bottom up that’s all.

⇒ Best thing is to post on X & Linkedin, i got some role offers just because they were impressed by my work, i think if you know things up to LLMs, RAG & little more, you’re pretty much sorted. I am not saying you will get one as there is a lot of competition, but once you’ll build your presence and showcase ur work, ppl will dm you this is the best way. TBH i never did or got one internship but i do post my work & also recently applied to cohere labs, you’ll never know when you’re enough capable but yeah at least finish the 3 books HOLLM, AI eng. & LLM eng. handbook, then you can apply also build the projects introduced in the books & once you’ll read these books, you’ll gain a lot of idea what to build next.

Hope i made it clear bro :)

⇒ See tensor flow is kind of an old man now, in research & in lot of places pytorch is used, also it is easy for someone to jump from python to pytorch, though i never felt that tensor flow is hard, maybe because i didn’t explore it fully. Even karpathy sensei said good thing about pytorch, so yeah no disadvantages. Also one imp point is that HOML uses tensor flow so i recommend you all to go with the book & learn that, having tf in our skill list will help you for sure. And as i said it’s not hard. Also in dec (please check the date o’reilly media), the HOML publisher are coming up with pytorch so you guys can learn numpy, pandas & sklearn till the time.

⇒ By classical ml, i hope you mean like random forest, KNN, GNB, regression now this is first part of HOML book, the second part introduces Deep learning which includes Neural nets, CNN, RNN, GAN, RL. There is no point in jumping to MLops without DL.

Look all modern architectures like GPT, BERT, Llama uses Deep learning, neural networks are the core, so i recommend you go through the book HOML religiously, follow the book sequence it is good. Then HOLLM, then AI eng. then LLM eng. handbook & then explore MLops. First get the concepts done, then move to MLops.

Also you can add design ml systems but at least get the 3 books done then think of MLops.