Senior Data Scientist – Fintech (Credit and Fraud)

Deep analysis. Smarter systems. Real users. As a Senior Data Scientist, you’ll own credit and fraud risk problems end to end — from framing an ambiguous question, through building and validating the model, to getting it live in a real-time decision engine and defending it in front of a risk committee. Your models decide who gets credit and which transactions get stopped, for millions of customers and merchants across our markets.
The Fintech Data Science team sits at the centre of how Delivery Hero extends credit and protects the business from fraud losses, spanning underwriting, portfolio risk management, and fraud prevention across our products. We build the models and decision systems behind credit approval, limit management, and fraud detection, so the business can grow lending responsibly while keeping losses, chargebacks, and financial crime exposure under control. We work with AI every day — coding agents and LLM tools for pipelines, data exploration, code and test writing, and documentation — and we check what they produce rather than trusting it.
Be part of building the financial backbone of Delivery Hero. You’ll develop products that empower millions of customers and merchants, from seamless payments to innovative financial solutions like wallets and credit. Your work will support our path to profitability by creating financial flexibility for users and enabling smooth transactions across our markets.
Own problems end to end, from an unclear business question through data, modelling, deployment, monitoring, and adoption.
Build and ship models: fraud and anomaly detection, real-time transaction decisioning, acquisition scorecards, and behavioural limit management.
Give Risk, Compliance, Finance, Payments, and leadership a clear recommendation they can act on, and the reasoning behind it.
Work closely with Engineering on reliability, owning the full data science lifecycle including deployment.
Build the data foundations: data marts and pipelines that produce features and training datasets.
Drive experimentation through to model rollout, then monitor what’s live and fix drift at the source.
A track record of owning models end to end, from an unclear business question through build, validation, deployment, and real use by the business.
Full-stack breadth: SQL and data analysis (e.g. BigQuery), orchestration and scheduled feature and training-data pipelines (e.g. Airflow), modelling, deployment (e.g. Metaflow, Vertex AI), and production monitoring. Strong Python, testing, and CI/CD.
Fluency with AI tooling. You build with coding agents and LLMs every day, automate your own routine work, and spend LLM budget sensibly — the right model for the task, tight context, no tokens burned on what a script should do. And you catch what the output gets wrong.
Independence and speed. You pick the approach, cut the scope, and ship, then explain why.
Judgment under ambiguity. You can make a defensible recommendation when trade-offs have no single right answer: false positives vs. fraud caught, loss vs. approval rate.
Clear communication. You can explain modelling trade-offs to non-technical stakeholders and support your recommendation with evidence when it’s challenged.
Nice to have
Experience in fraud, credit risk, payments, or fintech.
Real-time or streaming decisioning systems.
Building and deploying deep learning models in production.
Building LLM- or agent-based solutions in production.

Employment type
Full-time

Work arrangement
Hybrid