Internship – Data Science (Customer AI & Communication Optimization)
The project
At ING Belgium, we've built Kraken, an AI system that helps determine which messages are most relevant for each customer. Today, Kraken powers the personalised banners customers see in our app and online banking channels, helping us make communications more useful while avoiding unnecessary noise.
We're now looking at a different challenge.
Instead of deciding what to show when a customer happens to open the app, we want to understand when it's worth reaching out proactively. Which customers should receive a communication? Which channel works best? Is there a better moment to contact them? And just as importantly, when should we not send anything?
That's the question this internship will explore.
What you'll work on
Over three months, you'll help investigate whether the principles behind Kraken can be applied to outbound communications.
The project starts with understanding how customer communications are managed today and identifying where data science could improve decision-making. From there, you'll explore the available data, test different approaches, and evaluate whether they create measurable value.
Typical activities include:
- Exploring customer and communication data
- Understanding how Kraken was designed and what can be reused
- Framing a business question as a data science problem
- Comparing different modelling or decision-making approaches
- Running historical analyses and simulations
- Assessing opportunities for experimentation and real-world validation
The final objective is a proof of concept that demonstrates clear business impact and could potentially evolve into a larger production initiative.
Why this project is interesting
Most internship projects start with a predefined problem and a clear roadmap. This one doesn't.
We're entering a space where there are still open questions and genuine decisions to make. Some of Kraken's ideas may transfer surprisingly well. Others may turn out to be unsuitable for outbound communication. Part of the challenge is figuring out where those boundaries are.
You'll be working on a real problem with real customers in mind, and your conclusions can directly influence how future communication strategies are designed.
Team and support
You'll work closely with two Data Scientists who were involved in building Kraken from the ground up. Along the way, you'll collaborate with Data Analysts, Machine Learning Engineers, and other specialists working across customer analytics and AI.
While you'll own a substantial part of the project, you won't be left on your own. Expect regular discussions, technical guidance, and plenty of opportunities to challenge ideas together.
Technology
You'll work with:
- Google Cloud Platform (GCP)
- BigQuery
- BigQuery ML (BQML)
- Large-scale customer and communication datasets
Who we're looking for
This internship is well suited to students interested in data science, machine learning, analytics, or applied statistics.
Previous banking experience is not required. We're much more interested in how you approach problems than in your knowledge of financial products. Curiosity, critical thinking, and the ability to bring structure to ambiguous questions will take you a long way.
Useful skills include:
- Python and/or SQL
- Statistics, machine learning, or experimentation methods
- Strong analytical thinking
- The ability to work independently and take ownership
- Fluent English
Practical information
- Duration: 3 months
- Start date: November
- Full-time internship
- Hybrid working model – half-time in our Brussels office
What you'll gain
By the end of the internship, you'll have worked on a business problem from initial exploration through to evaluation and recommendation.
You'll also gain:
- Experience working with a production AI system operating at scale
- Exposure to the full data science lifecycle
- Coaching from experienced Data Scientists
- Collaboration with Data Analysts and Machine Learning Engineers
- The opportunity to contribute to a project with visible business relevance and potential customer impact