Expert Data Modeler, Fraud Risk Detection

Overview
Experian's Fraud Analytics & Commercialization operates across four main functions. These include client engagement analytics, scalable and custom analytics for financial institutions, fraud analytics consulting, and solution integrity and enablement for production-ready platforms.
We're looking for a motivated Data Scientist to help build fraud detection models and features that identify high-risk activity while minimizing friction for legitimate customers. Core skills for this role include an eagerness to collaborate, and empathy. You will will dig into surprising signals in the data and to learn how that insight becomes a deployed model.
You will help investigate the latest fraud patterns, build features, and train and evaluate machine learning models. You will work with senior data scientists and engineers starting with problem definition through feature engineering, experimentation, and deployment. You will be a developing programmer, ready to translate theoretical principles into production-ready solutions.
We continue to sharpen through research and the engineering that turns those findings into tools and systems built for commercialization.
This is a remote role and you will report into the Sr. Manager of Fraud Analytics.
What you'll do
Investigate large datasets, including exploratory analysis and fraud label development, to identify latest fraud patterns, attack methods, and behavioral signals.
Translate ambiguous fraud and risk problems into clear hypotheses, analytical plans, model requirements, and measurable success criteria.
Develop machine learning models for fraud detection across account opening, account takeover, and identity risk.
Evaluate models using metrics like ROC/AUC/KS/Gini, precision/recall, fraud capture rate, false-positive rate, customer friction, and fraud losses prevented.
Develop and validate predictive features using identity, transactional, consumer credit history, device, behavioral, temporal, velocity, network, and third-party data.
Write clean, efficient, well-tested Python and PySpark code, and collaborate with teams to bring models and features into batch, retro, or real-time decisioning environments.
Monitor feature quality, model performance, population changes, and fraud-pattern drift
Design and present analyses for model behavior, tradeoffs, risks, and recommendations
Follow appropriate standards for data privacy, model documentation, explainability, validation, and governance.
Qualifications
At least 3 years of experience in data science, machine learning, statistical modeling, or a related quantitative field
Bachelor's or advanced degree in computer science, statistics, engineering, data science, or another quantitative discipline
Direct experience developing fraud-detection, identity-risk, credit-risk, financial-crime, or other adversarial risk models.
Demonstrated experience creating meaningful fraud features
Proficiency in Python and PySpark, with experience writing modular and tested code for large datasets and distributed or cloud data systems.
Experience using common data science and machine-learning tools such as pandas, NumPy, scikit-learn, XGBoost, TensorFlow, or comparable technologies.
Knowledge of supervised learning, model evaluation, feature selection, statistical inference, experimentation, and model calibration.
Experience navigating challenges common to fraud modeling, including class imbalance, delayed or incomplete labels, changing attack patterns, and model drift.
Experience moving models beyond experimentation and into production, either directly or in close partnership with engineering teams.
#LI-Remote

Employment type
Full-time

Work arrangement
Yes