My Honest Advice for Aspiring Machine Learning Engineers

want to be machine learning engineers.

I get it.

It’s a great job, with interesting work, great pay, and overall, it’s very cool.

However, it’s definitely not a walk in the park to become one. In this article, I aim to offer my unfiltered and candid advice to aspiring machine learning engineers.

This will be more of a pep talk, providing you with clear expectations of what it takes to become a machine learning engineer and whether it’s something you really want to pursue.

Learn every week

If you want to become a machine learning engineer, then you need to dedicate at least 10 hours each week to studying outside of your everyday responsibilities.

I am sorry if that upsets you, but again, if you want to land a job in the highest-paying tech profession, you need to put in more time and effort than other people. There is simply no way around it.

Without sounding arrogant, I learn something new in machine learning every single week, even though I work full-time, create YouTube videos, exercise five times a week, and have mentoring and coaching clients. If I can make time, so can you. It all comes down to priorities.

Almost everything I’ve achieved in my career comes from consistently studying and documenting my learning outside of work. I’ve written over 150 technical articles on Medium on topics such as:

…and many more. You can see the complete list here.

This isn’t to boast but to show the level of commitment required to become a machine learning engineer.

Think of this profession in the same category as lawyers, doctors, or accountants. These fields demand years of study and practice. The same is true for machine learning; it’s not often seen as that due to its relative newness.

I often say:

Everything is easy, but hard.

It’s easy to understand what you need to do but hard to do it consistently over time. There is no secret; you have to take the long road.

So, pick something you want to learn and stick to it until the end; then, recycle this process again and again. That’s all there is to it.

Extend your time horizon

Even with the most ideal background, it will still likely take at least two years to become a fully qualified machine learning engineer at a top company.

Don’t fall into the trap of thinking that a few online courses and projects are enough to land a job in one of today’s highest-paying tech roles.

Online certifications help you learn the content in data science and machine learning, which is very valuable. However, they rarely allow you to get hired nowadays, especially in our tough job market.

I don’t say this to discourage you but to set realistic expectations. I’ve spoken with many people who try to shortcut their journey, and I’ve yet to see it succeed.

To become a machine learning engineer, you need solid foundations in:

Mathematics
Statistics
Machine Learning
Software Engineering
DevOps
Cloud Systems

Some of these skills can only be developed through real-world experience. That’s why I usually recommend people start as data scientists or software engineers first and then pivot to machine learning engineers, as it’s not an entry-level role.

Accepting the fact that it will take you a few years to become a machine learning engineer is liberating and takes the pressure off you.

Take your time to learn things deeply, really study, and your knowledge will build over time. I promise, eventually, you’ll be ready for that ML engineering role when the time is right.

Stop chasing AI

Newsflash: A machine learning engineer is not an AI engineer. So stop thinking that calling a chatbot API like ChatGPT or Claude makes you a machine learning engineer.

As a machine learning engineer, you’re expected to deeply understand how models/algorithms work and have a firm grasp of statistical learning theory and all the fundamental mathematics.

That means knowing core algorithms like:

Inside and out.

Most people claim to know them, but you’ll be surprised at how little you actually know.

I’ve mock-interviewed countless candidates, and many can’t even explain gradient descent from first principles using calculus.

Again, I’m not trying to be harsh but to show you the reality I have seen.

I always tell people to stop rushing to learn flashy topics like NLP, computer vision, or generative AI.

Your first few years should be about mastering the fundamentals; mastering them thoroughly so you have a solid understanding for many machine learning theory interview.

The reality is that most machine learning engineer roles primarily focus on classical supervised learning. Your job is often less about building exotic models and more about tailoring well-understood algorithms to solve specific problems. That’s why a deep understanding of the basics is essential.

If you want to test your fundamental knowledge, I offer mock interviews based on real questions I’ve faced in actual ML job interviews. Feel free to check it below.

Mock Interview with Egor Howell
Customised for your particular role and interviewtopmate.io

It is very hard

Let’s end with something that might seem a bit obvious: becoming a machine learning engineer is just hard.

As I’ve said throughout this post, the role demands expertise across a wide range of disciplines. You’ll need strong foundations in maths, statistics, and programming, plus real-world experience as a software engineer or data scientist first (which are tough jobs in their own right). Additionally, you must commit to continuous learning throughout this entire period.

Even with the most perfect background — a STEM master’s or PhD — it’s still a long, difficult journey. If you’re coming from a non-traditional path, it’s even harder. That doesn’t mean it’s impossible, but it is more difficult, and you need to decide if the challenge is worth it for you.

I often say:

Anyone can become a machine learning engineer — but that doesn’t mean everyone should, or even wants to.

It takes sustained effort for at least a few years.

You have to be honest with yourself about whether you’re willing to invest 2–3 years minimum (and, in most cases, 4–5 years) to break into the field.

That’s a long time.

For me personally, giving up four years for a decades-long career doing work I love is absolutely worth it. But that’s a calculation only you can make.

In fact, I find it liberating that it’s so hard, as it makes me feel better about struggling through it.

I am someone who doesn’t sugarcoat anything, and you might have noticed that most of my points boil down to two key factors: time and effort.

Anything worth doing often requires consistent effort over a long period. That is the secret to becoming a machine learning engineer.

If you are serious about becoming a machine learning engineer, then I recommend checking out the below article, where I detail my roadmap:

Link.