AI is transforming the way companies operate, and almost every company is looking for ways to leverage this technology.
As a result, the demand for AI and machine learning skills has skyrocketed in recent years.
With nearly four years of experience in AI/ML, I have decided to create the ultimate guide to help you get into this rapidly growing field.
Why do you work for AI/ML?
It is no secret that AI and machine learning are some of the most desirable technologies these days.
Being familiar with these fields will open up many career opportunities in the future, not to mention you are at the forefront of scientific advancement.
And to dull you will be paid a lot.
According to levelfyithe median salary for machine learning engineers is £93,000, and £75,000 for AI engineers. On the other hand, for a data scientist, that's £70,000, and for a software engineer, it's £83k.
Please don't get me wrong. These are very high pay in themselves, but AI/ML gives you that advantage and the differences can become more pronounced in the future.
You also don't need a PhD in Computer Science, Mathematics, or Physics to work with AI/ML. Good engineering and problem-solving skills are sufficient to have a thorough understanding of basic ML concepts.
Most of the work is not research work, but rather more AI/ML solutions for real problems.
For example, I work as a machine learning engineer, but I have not done any research. I aim to use algorithms to apply them to business problems to benefit the customers and therefore the company.
Below is a job that uses AI/ML:
- Machine Learning Engineer
- AI Engineer
- Research Scientist
- Research Engineer
- Data Scientist
- Software Engineer (AI/ML Focus)
- Data Engineer (AI/ML Focus)
- Machine Learning Platform Engineer
- Applied Scientist
They all have different requirements and skills, so there is one that suits you.
If you want to learn more about the above roles, we recommend reading the previous article.
Do I need to become a data scientist, data analyst, or data engineer?
Explaining differences and requirements between different data rolesmedium.com
Yes, let's get on the roadmap now!
Mathematics
I argue that solid mathematics skills are probably the most important for any tech expert, especially when using AI/ML.
A good foundation is needed to understand how AI and ML models work under the hood. This will help you to debug them better and develop intuitions about how to work with them.
Please don't get me wrong. You do not need a doctorate in quantum physics, but you should be aware of three areas:
- Linear algebra – Understand how matrices, eigenvalues, and vectors used everywhere in AI and machine learning work.
- Calculus– Understand how AI actually learns using algorithms such as backpropagation that utilize gradient descent and integration.
- statistics – Understand the probabilistic properties of machine learning models through learning probability distributions, statistical inference, and Bayesian statistics.
resource:
This is almost everything you need. If anything, that's a bit overloaded in some respects!
Timeline: Depending on the background, this should take months or months to speed up.
Here is a detailed breakdown of the mathematics required for data science. This can be applied to AI/ML as well.
Python
Python is the gold standard for machine learning and AI and is a reliable programming language.
Beginners get caught up in the so-called “best way” to learn Python. They teach the same things, so either introductory course is sufficient.
The main things you want to learn are:
- Native data structures (dictionaries, lists, sets, and tuples)
- For the loop
- IF-ELSE Condition Statement
- Functions and classes
I would also like to learn about specific scientific computing libraries such as:
- numpy – Numerical calculations and arrays.
- Panda – Manipulating and analyzing data.
- matplotlib & conspiracy– Data visualization.
- Scikit-Learn – Implementation of classic ML algorithms.
resource:
Timeline:Again, depending on the background, this should take several months. If you already know Python, it's much faster.
Data Structures and Algorithms
This may seem slightly out of place, but if you want to be a machine learning or AI engineer, you need to know the data structures and algorithms.
This is not just an interview. It is also used in AI/ML algorithms. You'll come across more than you think in backtracking, depth initial search, binary trees, and more.
Here's what you need to learn:
- Arrays and Link Lists
- Trees and graphs
- Hashmap, cue, stack
- Sorting and search algorithms
- Dynamic programming
resource:
- neetcode.io – Excellent introduction, intermediate and advanced data structures and algorithm courses.
- Leetcode& Hakke rank – A platform to practice.
Timeline:It took about a month to nail the basics.
Machine Learning
This is the beginning of the fun!
The previous four steps involved preparing a foundation to tackle machine learning.
Machine learning generally falls into two categories.
- Monitored learning– There is a target label for training the model.
- Unsupervised learning– If there is no target label.
The diagram below shows this division of each category and some algorithms.
The important algorithms and concepts to learn are:
- Linear, logistic, polynomial regression.
- Decision-making trees, random forests, trees added to gradients.
- Supports vector machines.
- K-means and K-nearest Neighbor Clustering.
- Functional engineering.
- Rating metrics.
- Regularization, trade-offs between bias and variance and cross-validation.
resource:
Timeline:This section is very dense and can take about three months to get to know most of this information. In fact, it takes years to truly master all of those resources.
AI and deep learning
Since the release of ChatGpt in 2022, there has been a lot of hype around AI.
However, AI itself has been a concept for a long time, dating back to the 1950s in its current form. Neural networks have been born.
The AI we mention at this time is specifically called the Geneai AI. This is actually a very small subset of the entire AI ecosystem, as shown below.
As its name suggests, Genai is an algorithm that generates text, images, audio, and even code.
Until recently, the AI landscape was dominated by two major models.
However, in 2017, a paper was called “You need to be careful.”It was published and introduced a trans architecture and models. This is after CNN and RNN have been replaced.
Today, trances are the backbone of large-scale language models (LLM) and clearly dominates the AI landscape.
With all this in mind, here's what you need to know:
- Neural Networks–An algorithm that actually places AI/mL on the map.
- Convolutional and recurrent neural networks –It is still used quite a bit today for certain tasks.
- Trans –Current latest technology.
- RAG, Vector Database, LLM Fine Tuning –These technologies and concepts are important to today's AI infrastructure.
- Reinforcement learning– The third type of learning used to create AI Alphago.
resource:
- Deep learning specialization byAndrew NG. -This is a follow-up course from the specialization of machine learning and will teach you everything you need to know about deep learning, CNNS, and RNN.
- Introducing LLMS Andrej Karpathy (former senior director of AI at Tesla) –Learn more about LLMS and how to train them.
- Neural Network: Hero from scratch–It starts relatively late and builds neural networks from scratch. But in the final video, he will have you build your own generation pre-training transformer (GPT)!
- Reinforcement learning course – Lectures by David Silver, the lead researcher at Deepmind.
Timeline:There are a lot here, and it's very hard and cutting edge. So about three months is probably what will take you.
mlops
As I've said many times, the Jupyter notebook model is worthless.
For AI/ML models to be useful, you need to learn how to deploy them in production.
The areas you learn are as follows:
- Cloud technologies such as AWS, GCP, Azure.
- Docker and Kubernetes.
- How to write production code.
- git, circleci, bash/zsh.
resource:
- Practical mlops (Affiliate link)-This is probably the only book you need to understand how to deploy a machine learning model. I use it more as a reference text, but it tells you almost everything you need to know.
- Machine learning system design (Affiliate link)– Another great book and resource for changing your sources.
Research paper
AI is evolving rapidly, so it's worth keeping all the latest developments up to date.
Here are some papers I recommend reading:
See the comprehensive list here.
Conclusion
Intruding into AI/ML may seem overwhelming, but taking steps one at a time.
- Learn the basics like Python, mathematics, data structures, and algorithms.
- AI/ML Knowledge Learning Obtained Learning, Neural Networks, Transformers.
- Learn how to deploy AI algorithms.
The space is huge, so it'll take about a year to get a full grasp of everything on this roadmap. That's fine. Literally, there's a bachelor's degree specialized in this space, and it takes three years.
Just go at your own pace and ultimately you will get to where you want to be.
Happy learning!
One more!
Join my free newsletter, Data Plateshares weekly tips, insights, and advice from my experience as a practical data scientist. Plus, as a subscriber, you get me Free Data Science Resume Template!
Data Plate | Egor Howell | Subsac
Advice and learning about data science, technology and entrepreneurship. Click to read the data, Egorhower,Newsletter.egorhowell.com
