My Story of Becoming an MLOps Engineer
Here is my personal journey. I’ve been working in the MLOps field for 5 years, mostly as an ML Engineer. How did I even get here?
In my time, grades mattered, and a diploma was essential. It wasn’t really a time when you could take an online crash course in Python and immediately transition into a developer role. I was your typical nerd.
In 2012, I stepped into my first programming class (Java OOP) at university, to start my Computer Science & Engineering degree. The room was packed with a bunch of genius video gamer men and some genius women.
Java is a great programming language to start with. It teaches you the fundamentals of coding. And I hated it. I hated programming. So how was I going to finish 4 years of CSE degree and work in that field?
This was me. No exaggeration.
Luckily I was good at other engineering core courses — Maths, Physics, Chemistry, Economy, etc. In my 3rd and 4th years, we were required to select specialized courses.
The turning point. I remember the exact moment that got me here today. I was choosing between 2 courses, ML and Networking. I chose ML. Today, I am a proud ML engineer who can build end-to-end ML pipelines.. while my networking knowledge is still meh, I can’t even fix my wifi. Sorry to all the networking lovers.
That was the moment I met with Machine Learning. I had no clue before, and the world hasn’t been awakened with AI yet. There were only 11 students who were chosen to be in that class. (The professor was very tough). And I loved it. It was obviously full of maths and statistics. Building something that can be learned from data? Fantastic. Later, I continued with NLP. Then I worked on Computer Vision for my senior thesis about detecting logos and signatures in documents. I finally knew what I wanted to do, I applied to several master’s programs in Europe, but only one offered a full-focused AI curriculum. So I got into the AI Master program at the University of Amsterdam. 2 years of only AI-centric courses.
Then AI boomed. I remember the time when Google paper, Attention is All You Need, came out, which is the recipe behind Google Translate. My educational pivot was very much aligned with the industry trend. This part was more luck than me foreseeing things in my crystal ball.
During the second year of my master’s program, I started working as a data science intern at a company. And I really enjoyed it, possibly due to fantastic colleagues, free lunch and Friday drinks. Pursuing an academic career has always been a possibility for me. In fact, it was my first choice before I started the master’s program. However, working in a company during my master’s thesis brought a different perspective, and I realized that I enjoyed applying what I learned to real-world use cases more than engaging in research. Also, let’s be real — the financial aspect is hard to ignore. No shame in admitting that earning more money is appealing.
After graduation, I started working at ABN Amro, a Dutch bank, where I mostly did data science work, built NLP models, and engaged closely with business stakeholders.
Later I wanted to get closer to the engineering side and I switched to Ahold Delhaize, as an ML engineer. There, I focused on implementing solution designs for ML models, as well as deploying and maintaining them. Working within our team at Ahold Delhaize provided me with valuable experience in MLOps, product management, and speaking at conferences. My years of work at Ahold have contributed the most to my growth as an engineer, it also introduced me to Maria who is my co-founder of this blog.
Currently, I’m working at Booking.com in the ML platform team, building infrastructure, developing services, and sometimes coding in Java.
My education, which I started with a dislike for Java, and my career choice, which I chose to work in ML models rather than become a developer, ironically brought me to this day where I work by writing Java code and work more as a developer than a data scientist.
Joking aside, I enjoy having worked in slightly different roles, in different companies, and moving closer to the creation of infrastructure for ML services.
What I love about MLOps is that it is so inevitable, every company that started its AI journey by hiring data scientists is now looking into building a proper MLOps team. To bring ML models to life, the operational side is essential. What I like the least about MLOps, is probably the amount of tools being released every single day. Isn’t it getting a little overwhelming? Adding here the MAD landscape.
A little wisdom I picked from my previous manager at ABN Amro, is that you may not always know what you want in life but you always know what you don’t want. Think about it — if something gives you a discomfort feeling, not making progress, then why waste your time?
There are (still) enough job openings in this field for all of us.
Always focus on your own growth and learning. Never work “for” a company. Always work “for” yourself. Keep in mind that your career is your own path, and focusing on personal growth will eventually lead to more happiness and achievements. In the end, it’s not about the job title or the company you work for, but the fulfillment you find in following your own aspirations and dreams.
