Meet the Expert, Mike Huls

"You have to be a bit of a geek"

The world of data and AI is evolving at lightning speed. Organisations feel the urgency to do something with AI, but often struggle with where to start. And what do you actually need to use AI successfully? Mike Huls, Principal Expert Data & AI at Sopra Steria, helps clients translate organisation-wide challenges into scalable data and AI solutions. We spoke with Mike about his career, his vision on data & AI, and why you have to be a bit of a geek to thrive in this field. 

What does a Principal Expert Data & AI do?
"The role of a Principal Expert Data & AI is to translate organisation-wide challenges into scalable data and AI solutions. It’s never just about the technology. A good solution aligns people, processes, and technology with the organisation’s objectives. That sometimes means making difficult decisions and occasionally saying no, because you can’t meet every requirement, especially when it comes to security. And yes, that can lead to some tough discussions. Sometimes a client already has a particular solution in mind beforehand, but that usually isn’t the best approach. We first go all the way back to defining what the problem actually is. Once that’s clear, we look at the functional requirements. Then, based on the problem analysis and those requirements, we decide which technology is best suited to solve it."

How did you get into IT?
"Although I was fascinated by computers as a child and was always experimenting, I didn’t study IT. I studied Real Estate in Groningen because that’s what interested me at the time. However, in my first job in commercial real estate, I realised that I got most of my energy from automation and data, while the real estate sector was still fairly conservative in that area. Because I wanted to work more with data, I resigned and moved into business intelligence, focusing on data visualisation. But you can only do that properly if you know how to program. So I taught myself programming in the evenings and never stopped."

How long have you been working at Sopra Steria?
"I joined Sopra Steria in January 2025, so just over a year ago. In the beginning, I mainly helped build up the Data & AI practice in the Northern region. I already had experience with that, as I had done something similar at my previous employer. We started with a small team of four. The practice has now grown to nine people, and I work on various client assignments. What I find particularly interesting is the combination of data and AI. I really enjoy making processes more efficient and user-friendly through automation. AI can be extremely helpful in that, but only if there is a strong data foundation underneath it."

Can you give a practical example of your work?
"At the National Coordinator Groningen (NCG), which investigates and reinforces homes in the earthquake-affected region of Groningen, I am the coordinator of the AI Center. Because the AI Center is positioned at the heart of the organisation, my role is quite broad: I lead the AI centre, advise on ethics, privacy and governance, and also deal with architecture, code reviews and stakeholder management.

At the same time, in order to complete case files, we need to classify around 1.4 million documents in the short term. That presents a significant technical challenge, not only because we have to develop and train classification models, but also due to the limited data quality. In addition, there are many stakeholders involved who need to be aligned.

To make the asignment manageable, we started working in short iterations, delivering results every few weeks. This short-cycle way of working, experimenting, implementing and adjusting as we go reduces risks and ensures better alignment with the client."

Which project are you most proud of?
"At a previous employer, a marketing automation company with DPG as a major client, I set up a data department. It was a challenging project where I brought together a group of people around me and ultimately built a well-functioning team to make the data department a success. And it worked.

When I started there, the onboarding period took months. Everything ran on legacy technology. I introduced new tooling, improved processes and trained people. By the time I left, we had scaled up massively and were able to handle a multiple of the previous workload with only a fraction of the people. In addition, a new colleague could submit their first pull request after just three days. That’s a really great result and a strong example of bringing technology, processes and people together to contribute structurally to organisational goals."

What are the biggest challenges for clients in data & AI?
"Many organisations currently feel the pressure to do something with AI, because they’re afraid of being left behind. So the sense of urgency is definitely there, but they don’t always know exactly what they should do. You can want to do something with AI, but that’s only possible if the foundation is solid. Without that foundation, AI applications often require a major investment for very limited return. For me, that foundation consists of three elements: data quality and accessibility, governance, and literacy.

First of all, if your data is inaccessible, low quality or simply doesn’t exist, it becomes difficult to solve data-related challenges: it’s garbage in, garbage out. For AI applications, that effect is amplified even further.

Secondly, within an organisation you often see all sorts of fragmented initiatives without proper oversight. This creates risks in areas such as security and privacy. Nowadays, it’s very easy and accessible to come up with a solution using vibe coding. With good governance, you ensure that these types of initiatives can be organised and directed centrally, so you can continue to benefit from the efficiency of AI while avoiding the security risks that come with it.

Finally, the majority of AI projects fail — not only because of poor governance or data quality. Literacy also plays an important role. Without knowledge of AI or data, it’s difficult to assess whether AI will actually contribute to your organisational goals. You can see this in overly high expectations: AI is not a magic wand you can simply wave and suddenly all problems are solved."

What makes your work so enjoyable?
"Every week in the AI world is different. There are constantly new possibilities. What inspires me most is helping people make their work more enjoyable and efficient. Automation takes the robot out of the human, so people don’t have to do repetitive tasks and can instead focus on work that truly requires a human touch."

And IT is far more social than people think. There’s often a stereotypical image of IT professionals being tucked away in a dark basement, endlessly hammering away at a keyboard, but I actually find it to be a very social profession. You have to communicate, align with others, and brainstorm together. It’s about understanding what someone needs and translating that into a technical solution. Experience has taught me that investing time upfront in understanding the problem pays off tenfold during development."

How do you stay sharp?
"You have to be a bit of a geek. I experiment a lot with new technologies and try things myself to really understand them. For example, I recently built a web application that allows you to anonymise sensitive information in documents and images before uploading them to an LLM. I also write on my blog and on Towards Data Science about software engineering, machine learning, AI, system design and architecture. I strongly believe you only truly understand something once you can explain it simply. Blogging forces me to really get to the bottom of it."

What advice would you give someone interested in IT?
"Just start. Grab a laptop and experiment. That’s the only way to find out if you genuinely enjoy it and if you’re happy spending eight hours a day solving problems on a screen. For example, use ChatGPT to build a Python application that scrapes financial data. It’s incredibly accessible nowadays. But don’t let AI do all the thinking for you. Use it to spar with, not to replace your own reasoning. Ultimately, you need to discover whether you enjoy analysing and solving problems. That’s the core of the profession."