In the Author Spotlight series, TDS Editors chat with members of our community about their career path in data science, their writing, and their sources of inspiration. Today, we’re thrilled to share our conversation with Egor Howell.
Egor is a data scientist and machine learning engineer specialising in time series forecasting and combinatorial optimisation. He runs a content and coaching business helping people break into data science and machine learning, as well as teaching technical topics
Let’s start at the beginning: What sparked your initial interest in data science, especially since you didn’t follow the traditional CS degree or bootcamp route?
I can pretty much single-handedly attest my career to DeepMind’s AlphaGO documentary. It made me incredibly curious about machine learning and its potential to solve virtually any problem. After that, I was looking for careers that use machine learning, and naturally, a data scientist came up. So, from then on, I basically self-studied to become one!
You’ve written about doing more than 80 data science interviews. What were some key insights you gained from that experience, both about the hiring process and your own growth?
Interviewing is a skill and is very different from what you do on the job. It’s basically a game, and you have to learn how to play it, like pretty much anything in life.
The central insight is that you fundamentally have to prepare; I am shocked about how many times candidates don’t even really know what the business does!
Another key point people overlook is the soft skills and the intangibles. Unfortunately, suppose someone is very monotone and shy but knows a lot. In that case, they are less likely to get the job compared to someone charismatic, friendly, and, in general, who brings good energy.
And finally, make sure you don’t speak for more than 2 minutes at a time. I have interviewed people who talk and talk and talk. If you’ve noticed you’ve been talking for a while, say something like, “I can go into more detail if you like.” This way, the ball is in their court, and they can move the interview forward if they wish. Nothing is worse than someone who keeps on speaking, as it doesn’t allow the interviewer to ask all their questions. Plus, it’s a skill to be able to explain yourself concisely.
One of your more provocative articles is titled “STOP Building Useless ML Projects.” Why do you think so many portfolio projects miss the mark, and what makes a project truly impactful?
People are always looking for a shortcut and don’t want to spend time thinking about a good-quality project. Any impactful project is personal to you, solves a problem or answers a question that you want to know, and takes you at least a month to build.
There’s no secret; it’s more about the effort people don’t want to put in most of the time. In that specific post, I have a framework for people to follow if they want to find an impactful project for themselves.
You often write with a clear audience in mind: Career switchers, beginners, and aspiring ML professionals. How do you decide what to write about, and who are you hoping to help most?
At first, it was tough, but now I ask my audience or read the comments to see what people are looking for.
My goal is to help people break into the field, but I’m being brutally honest along the way and not sugarcoating anything.
In most of my posts, I don’t “promise anything,” and I actually often say how hard it is and it may not be the right job for everyone.
What’s something that surprised you when you started working full-time as a machine learning engineer—something you wish more people knew going in?
You spend a lot of time maintaining models and infrastructure as opposed to developing models. The job isn’t exciting 100% of the time.
You’ve published a lot of career advice—from job prep to how to make a DS portfolio stand out. How has writing regularly shaped your own thinking, or even your career path?
Writing is thinking, so the better you write, the better you will think. What people don’t tell you is that a lot of the job is writing; you write plans, documents, tickets, etc. This skill is crucial because if you can explain yourself clearly, that goes a long way in life.
What trends in machine learning or AI are you personally most excited—or skeptical—about right now? How are those trends shaping your focus or ambitions?
I am a big “hater” of AI. I think it’s overrated, and it’s definitely not going to take over any jobs, at least in the next five years. Personally, I’m not putting much effort into learning it, as I think it’s a “flash in the pan.” I’d rather focus on areas that have been around for decades, such as statistics, operations research, time series, etc.
For someone who feels stuck—maybe they’re in a data analyst role, or struggling to break into ML—what’s the most practical next step they could take today?
Take everything one step at a time, and don’t try to think too far ahead. First, focus on projects, then your resume, then on applications, then on interviews, and then on the offer negotiation.
There’s no point in focusing on interviews if you’re not getting any; your time would be better spent on your resume and projects. Having a single focus is how you make progress.
To learn more about Egor‘s work and stay up-to-date with his latest articles, follow him here on TDS, on YouTube, and on LinkedIn.
