I’m looking for a structured learning path for AI. I don’t just want to learn how to use existing models—I want to understand how to build and train them from the ground up.
I want to dive deep into AI concepts, including machine learning, deep learning, neural networks, transformers, large language models (LLMs), optimization, training techniques, and model deployment.
I’m currently enrolled in AI for Everyone, but I’m not sure what the next step should be after completing it. Could you recommend a structured roadmap or sequence of courses that takes me from beginner to advanced, eventually enabling me to build production-ready AI systems and even develop my own models?
I want to add one point before you pursue these specializations. I think you should first go for the Mathematics for Machine Learning and Data Science Specialization. This is because many machine learning algorithms and systems rely heavily on the complex mathematical concepts that underpin them.
I see, thank you, hadnt read it properly. I do have a math background and python experience. My main concern with the math for ml course here was it would be too introductory or not enough exercises. I wouldnt mind doing it first if its interesting and math focused. But you say it uses a lot of non standard terminology. Would you still recommend I just skip and go straight to the two(Machine Learning Specialization,Deep Learning Specialization) or is there some benefit to still doing it in my case. Thank you!!
Yes, if you already have math background, I agree with Tom that the best plan is to start with MLS and then take DLS. You will then see if you run into any troubles with the math that is involved.
The key thing is that you need to be very familiar with the fundamentals of Linear Algebra: vectors and matrices and operations on them, including dot products, matrix multiply, transpose and the like. You don’t need anything as sophisticated as knowing what an eigenvalue is, but you need to be very comfortable working with vectors and matrices. If you have that background, then you should have no problem with the math in MLS and DLS.
I see little practical value in the Math for Machine Learning courses.
The few useful nuggets it contains are covered much better by Andrew in the MLS courses.
The only significant math concepts are how to compute a mean and standard deviation, and the concept of what happens when you multiply a vector and a matrix. You don’t need to know the implementation details, because Python knows how to do that for you. You do need to understand what “compute the sum of the products of the elements” means.
As a beginner myself, this discussion has been helpful. I recently completed the first course of MLS, in between I figured that I needed to familiarize myself with Python programming. Having knowledge in Python Programming is such a great booster, I didn’t have that when I started but I paused MLS and delved into python programming when I attempted the first optional lab. It’s been an interesting journey so far. I am excited to begin the second course and see what knowledge awaits me.
That’s a great goal. I’d focus on building a strong foundation in Python, math (linear algebra, calculus, probability), and machine learning before diving into transformers and LLMs-it makes the advanced topics much easier to understand.
I would recommend either going for a basic math course to at least see if the maths required will be something you can understand. Have you done much maths at school, college?