need to be machine studying engineers.
I get it.
It’s an excellent job, with attention-grabbing work, nice pay, and total, it’s very cool.
Nevertheless, it’s positively not a stroll within the park to change into one. On this article, I goal to supply my unfiltered and candid recommendation to aspiring machine studying engineers.
This shall be extra of a pep speak, offering you with clear expectations of what it takes to change into a machine studying engineer and whether or not it’s one thing you actually need to pursue.
Be taught each week
If you wish to change into a machine studying engineer, then it is advisable dedicate at the least 10 hours every week to finding out exterior of your on a regular basis duties.
I’m sorry if that upsets you, however once more, if you wish to land a job within the highest-paying tech career, it is advisable put in additional effort and time than different folks. There’s merely no manner round it.
With out sounding smug, I be taught one thing new in machine studying each single week, regardless that I work full-time, create YouTube movies, train 5 occasions per week, and have mentoring and training shoppers. If I could make time, so are you able to. All of it comes all the way down to priorities.
Nearly all the pieces I’ve achieved in my profession comes from persistently finding out and documenting my studying exterior of labor. I’ve written over 150 technical articles on Medium on matters equivalent to:
…and lots of extra. You’ll be able to see the entire listing here.
This isn’t to boast however to indicate the extent of dedication required to change into a machine studying engineer.
Consider this career in the identical class as attorneys, docs, or accountants. These fields demand years of research and apply. The identical is true for machine studying; it’s not usually seen as that as a result of its relative newness.
I usually say:
All the things is straightforward, however onerous.
It’s simple to know what it is advisable do however onerous to do it persistently over time. There isn’t any secret; it’s a must to take the lengthy street.
So, choose one thing you need to be taught and follow it till the top; then, recycle this course of repeatedly. That’s all there’s to it.
Lengthen your time horizon
Even with probably the most best background, it’ll nonetheless seemingly take at the least two years to change into a completely certified machine studying engineer at a high firm.
Don’t fall into the lure of considering that a number of on-line programs and initiatives are sufficient to land a job in one among in the present day’s highest-paying tech roles.
On-line certifications assist you be taught the content material in information science and machine studying, which may be very beneficial. Nevertheless, they hardly ever permit you to get employed these days, particularly in our tough job market.
I don’t say this to discourage you however to set sensible expectations. I’ve spoken with many individuals who attempt to shortcut their journey, and I’ve but to see it succeed.
To change into a machine studying engineer, you want strong foundations in:
- Arithmetic
- Statistics
- Machine Studying
- Software program Engineering
- DevOps
- Cloud Methods
A few of these expertise can solely be developed by real-world expertise. That’s why I normally advocate folks begin as information scientists or software program engineers first after which pivot to machine studying engineers, because it’s not an entry-level function.
Accepting the truth that it’ll take you a number of years to change into a machine studying engineer is liberating and takes the stress off you.
Take your time to be taught issues deeply, actually research, and your data will construct over time. I promise, finally, you’ll be prepared for that ML engineering function when the time is correct.
Cease chasing AI
Newsflash: A machine studying engineer is not an AI engineer. So cease considering that calling a chatbot API like ChatGPT or Claude makes you a machine studying engineer.
As a machine studying engineer, you’re anticipated to deeply perceive how fashions/algorithms work and have a agency grasp of statistical studying concept and all the elemental arithmetic.
Which means realizing core algorithms like:
In and out.
Most individuals declare to know them, however you’ll be stunned at how little you really know.
I’ve mock-interviewed numerous candidates, and lots of can’t even clarify gradient descent from first ideas utilizing calculus.
Once more, I’m not making an attempt to be harsh however to indicate you the fact I’ve seen.
I at all times inform folks to cease dashing to be taught flashy matters like NLP, laptop imaginative and prescient, or generative AI.
Your first few years ought to be about mastering the basics; mastering them completely so you may have a strong understanding for a lot of machine studying concept interview.
The truth is that the majority machine studying engineer roles primarily deal with classical supervised studying. Your job is usually much less about constructing unique fashions and extra about tailoring well-understood algorithms to unravel particular issues. That’s why a deep understanding of the fundamentals is important.
If you wish to check your basic data, I provide mock interviews based mostly on actual questions I’ve confronted in precise ML job interviews. Be happy to verify it beneath.
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.
One other factor!
I provide 1:1 teaching calls the place we will chat about no matter you want — whether or not it’s initiatives, profession recommendation, or simply determining the next step. I’m right here that will help you transfer ahead!
1:1 Mentoring Call with Egor Howell
Career guidance, job advice, project help, resume reviewtopmate.io
