Team description The Card Not Present (CNP) Protect team is a cross-functional group within SumUp's Risk & Compliance tribe, responsible for keeping millions of merchants safe by preventing fraud across our card-not-present products.
We build production-grade ML systems and data pipelines that power real-time automated decisioning — and we're now expanding our infrastructure to support foundation model embeddings.
This is a high-ownership, high-impact role at the core of SumUp's financial safety mission, where the quality of your data work directly shapes how well our fraud detection performs.
👉 Take a look inside our Berlin.
What you'll do Design, build, and maintain production-grade Python-based data pipelines that power ML workflows, including real-time and near-real-time data processing Take full ownership of data quality and reliability — implementing validations, automated testing, monitoring, and alerting with clearly defined SLAs Strengthen our data foundations by documenting architecture, data lineage, dataset definitions, and dependency management Lead Feature Store governance, improve usability, and standardise our feature store setup so data scientists can move faster and with confidence Work closely with the Risk Platform, ML Data Platform, data scientists, software engineers, and analysts to deliver changes safely to production  You'll be great for this role if Strong proficiency in Python and PySpark, with hands-on experience designing computationally efficient solutions in large-scale production environments Experience building and maintaining feature engineering pipelines, including enriched attributes for online and offline use cases Comfort working with cloud infrastructure such as AWS (S3, EKS, Keyspaces, Athena) or equivalent providers, alongside containerisation tools like Docker and version control with Git Experience with streaming or event-driven architectures (such as Kafka) and familiarity with open table formats like Apache Iceberg.