As there are quite a few, I am going to break them down into the following categories:
- Programming and software engineering
- Maths and statistics
- Machine learning
- Deep learning and LLMs
- AI engineering
Programming and software engineering
If you want to work in AI, you must learn to program and have good software engineering skills.
As the field is relatively new, the de facto languages for AI are still up in the air. However, Python is your best bet to learn because of its ease of use and AI infrastructure.
AI jobs have mainly been spun up from machine learning, where the lingua franca is Python, and this is not changing anytime soon.
However, the most popular AI role, AI engineer, is closer to software engineering than machine learning engineering, so you may need to learn other backend languages like Java, GO or Rust.
I recommend starting with Python as it’s much easier and lets you understand the key software engineering fundamentals, but you may have to pivot languages in the future.
Although there are many courses and books, the best teacher is consistent practice. While resources will help you start your journey, creating and building is how you will really learn Python and, in fact, any language.
My main recommendations for Python and software engineering fundamentals are:
- Learn Python — Full Course for Beginners — The first course I took on Python at the start of my journey. It’s only 4 hours long, so you can do it in half a day.
- Python for Everybody Specialization — This is probably the most recommended course out there, and for good reason. If you are after an end-to-end course to learn Python, then this is it. Any reputable “Intro to Python” course will suffice, though.
- Hacker Rank & Leetcode — I used this when prepping for Python coding interviews.
- NeetCode — I used this resource to learn about data structures, algorithms, and system design. It’s an excellent platform for learning all the basic and advanced topics with hands-on exercises and delivers great interview preparation.
- Harvard CS50 Introduction to Computer Science — If you have been anywhere in the online tech space, you would have heard of this course. It’s probably the best intro to computer science and software engineering course! Highly recommend it to a complete beginner and, in fact, anyone.
Maths and statistics
Even though you may argue that you don’t need to know the maths, as most AI jobs are mainly about implementing foundational models, if you want to be a top AI practitioner, you should know at least how these models work under the hood.
The following resources are all you need to learn the required maths; I don’t think you need to look elsewhere.
- Practical Statistics for Data Science (affiliate link)— This would be it if you could get only one book to learn statistics. The main draw is that it provides statistics knowledge specifically for AI/ML practitioners, with hands-on examples in Python.
- Mathematics for Machine Learning (affiliate link)— This is a comprehensive book on the maths behind machine learning and AI, covering topics like calculus and linear algebra. It is pretty advanced, so I don’t recommend going through the whole thing end-to-end. Instead, use it to learn key concepts and as a reference text.
- Mathematics for Machine Learning and Data Science Specialization — This is a newly released course by DeepLearning.AI, the makers of the famous Machine Learning and Deep Learning specialisations. It’s ideal for beginners and covers all the fundamental maths topics, such as calculus, linear algebra, statistics, and probability, relevant to AI and machine learning specifically.
Machine learning
The majority of current AI actually refers to GenAI, a subsection of machine learning. As its name suggests, GenAI are algorithms that generate text, images, audio, and even code.
However, AI has been around as a concept for a long time, dating back to the 1950s, when the neural network originated.
It even predates that, with Alan Turing coining the “Turing Test” after his work on computers and thinking machines during the Second World War.
Anyway, my point is that AI is so much broader than most people think today, and you need a solid grounding in machine learning and traditional AI to be a great current day AI professional.
The following list will cover all your baseline machine learning knowledge; if you want to learn more advanced topics like time series forecasting, reinforcement learning, optimisation or computer vision, let me know, and I can recommend you some.
- Hands-On ML with Scikit-Learn, Keras, and TensorFlow (affiliate link) — If I could only give you one book to help you learn machine learning and AI, it would be this. It is fantastic, covers almost everything you need to know, and even touches upon LLMs, reinforcement learning and computer vision right at the end.
- Machine Learning Specialization — The first course I took on machine learning back in 2020 and is probably the best course on machine learning in history. When I took it, it was in Octave, but it has since been revamped, is now in Python, and has more cutting-edge topics like recommender systems and reinforcement learning.
- The Hundred-Page ML Book (affiliate link) — All machine learning is summarised in 100 pages! Really nice reference text for looking up things quickly and getting a refresher. Covers the basics really well.
- The Elements of Statistical Learning (affiliate link) — Excellent for mastering machine learning fundamentals, basically statistical learning. This book will truly teach the essence of machine learning.
Deep Learning and LLMs
As I showed in the diagram above, deep learning is a smaller category within the overall AI umbrella and a subsection of machine learning.
Deep learning is where all these generative AI algorithms came from, so you will truly study how LLMs, diffusion, transformers and all the other foundational models work under the hood.
- PyTorch Tutorials — You would need to learn a deep learning framework, and there are two libraries leading the industry: PyTorch and TensorFlow. I would focus on learning PyTorch as it is becoming increasingly popular compared to TensorFlow. PyTorch was used in about ~77% of research papers in 2021, and ~92% of models are PyTorch exclusive on HuggingFace.
- Deep Learning Specialization — Follow-up to the machine learning specialisation course. It’s a great course for learning Convolutional Neueral Networks and Recurrent Neural Networks and an intro to LLMs. Probably the best deep learning course out there!
- Introduction to LLMs by Andrej Karpathy (former senior director of AI at Tesla) — An hour intro video to LLMs and how they are used. I recommend this to anyone, even if they don’t plan on studying AI. It is very high level and not technical at all.
- Neural Networks: Zero to Hero (also by Andrej Karpathy) — Starts relatively slow, building a neural network from scratch. However, in the last video, he gets you building your own Generative Pre-trained Transformers (GPT)! A must tutorial for anyone serious about working as an AI practioner.
- Hands-On Large Language Models: Language Understanding and Generation (affiliate link) — There aren’t many textbooks on LLMs, as they have only gone mainstream in recent years. However, I have heard great things about this one. It was co-written by Jay Alammar, who wrote probably the best blog post explaining transformers ever and is endorsed by Andrew Ng.
AI Engineering
At this point, you will thoroughly understand the AI landscape, particularly LLMs and GenAI models, both hands-on and theoretically.
The real value comes from creating products from your AI models and knowledge. Therefore, you need to learn how to productionise and deploy these algorithms so they can benefit customers and businesses.
Most AI jobs are so-called AI engineers, and it’s closer to traditional software engineering than machine learning engineer jobs.
It’s mostly about using foundational GenAI models like LLama, GPT-4, and Claude and building products around them. You rarely do actual model development, mainly because training these models is expensive, and the current foundational models are so good!
- Practical MLOps (affiliate link) — This is probably the only book you need to understand how to deploy your machine-learning and AI models. I use it more as a reference text, but it teaches almost everything you need to know, like containerisation, shell scripting, cloud systems and model monitoring.
- AI Engineering (affiliate link) — This book is very popular at the moment. It’s written by Chip Huyen, who is arguably the leading expert behind ML/AI systems in production. She even taught a course on it at Stanford! Therefore, you are in good hands by using this book.
There are tons of resources; the main point is to not overcomplicate and start. They all teach the same things roughly, so you won’t go wrong no matter what course or book you use.
Another thing!
I offer 1:1 coaching calls where we can chat about whatever you need — whether it’s projects, career advice, or just figuring out your next step. I’m here to help you move forward!
1:1 Mentoring Call with Egor Howell
Career guidance, job advice, project help, resume reviewtopmate.io
