Role · Data scientist
When do you need a data scientist?
You need a data scientist once a fixed rule or a calculation no longer does the job and you want a model that learns from what happened before. Building that model is the easy part. The difference is whether they think about maintenance and explainability from day one.
What do you do in this role?
Models and predictions are the easy part. A good data scientist thinks from day one about how the model will be maintained and explained.
- Developing and validating models
- Making predictions usable
- From proof of concept to production
What companies look for in a data scientist
- Someone who first asks whether the problem needs a model at all. Often the answer is no, and saying so out loud saves months.
- Someone who can explain a model to whoever decides on it: certainly when that decision is about people.
- Attention to what comes after the pilot: maintenance, retraining, and noticing when a model quietly gets worse.
- Statistics that hold up. A good score on the test set means nothing if the set-up leaks.
What matters
- Python, with the usual toolkit, and SQL to get at the data.
- Statistics: not only applying it but knowing when an assumption does not hold.
- Being able to validate. Anyone without a sound validation design is marking their own homework.
- Enough engineering to get a model out of a notebook.
- Explainability, and knowing where it is legally required.
How the role is changing
Building a model has become cheap; maintaining one has not. The question therefore shifts from "can you build a model" to "can you keep one in production and account for it". On top of that, generative AI eats part of the classic work: what used to need a bespoke model is sometimes a well-framed question to an existing one. Those who grow here move towards ML engineering, or towards the problem itself.
How you stand out among the other hundred
By showing one project that reached production and stayed there. Not the Kaggle score but the question of who uses it now, how often, and what happened when reality changed. That separates someone who builds models from someone who solves something with them, and the second is what organisations are looking for.
What every role has in common
Companies increasingly look for someone who is broadly deployable. That does not mean you have to be able to do everything: it means you look into the disciplines around yours.
As a BI specialist you do not have to be an AI specialist. But if you demonstrably use AI in your work and keep learning about it, you become added value rather than replaceable. The same goes for communicating across departments: anyone who recognises the problems, pitfalls and opportunities of sales or finance delivers work that actually lands there.
What you can earn
Our own rates are based on a gross full-time monthly salary of € 3.000 to € 3.200. That is the floor we work from for a starting professional working full time.
A market indication per role is deliberately absent. The same job title varies widely between companies (a senior BI specialist earns noticeably less at one employer than at another) and the region weighs heavily: Amsterdam sits above Eindhoven, Eindhoven above Rotterdam, and Rotterdam above Drenthe. Quoting a number without checking it would mislead you in a conversation about your own income.
Just ask us. We see what the market pays and we will tell you what we know, including when we do not know.
Current vacancies
There is no vacancy open in this role right now. Sign up anyway: most of the work we do is never advertised.
Through which brand does a data scientist reach you?
- New Data Student
- a student or starter, with guidance, for the work that keeps being postponed
- New Data Squad
- an experienced professional who sets the lines and guides the change
- New Data Search
- somebody you employ yourself
Which form fits depends on the question, not on what we happen to sell.
Do you work as a data scientist?
Leave your details. We find the assignment that fits what you can do, not the other way round.