Machine Learning Engineer — Recommendation & Fraud

About the role

We are building machine learning systems that power two of our most critical product capabilities: personalised financial product recommendations and real-time fraud detection. We need an ML Engineer who can take models from research through to reliable production deployment and who understands that a model's value is entirely determined by what it does in the real world. Our recommendation engine suggests savings products, insurance plans, and investment portfolios based on a customer's transaction history, demographic signals, and behavioural patterns. Our fraud detection system processes every transaction in real time, scoring it for anomaly signals and triggering step-up authentication or blocking when confidence thresholds are breached. Both systems must be fast, explainable, and fair — we take seriously the risk of algorithmic bias in financial services. You will own the full ML lifecycle for these systems: feature engineering from our data warehouse, model training and evaluation, A/B testing frameworks, model serving via FastAPI, monitoring for data drift and model degradation, and retraining pipelines. You will work closely with our data engineering team to ensure the features you need are available in the feature store, and you will work with product managers to define metrics that accurately capture business impact. You bring strong Python skills, deep familiarity with scikit-learn, XGBoost, and at least one deep learning framework (PyTorch preferred). You understand how to evaluate classification models for imbalanced datasets, which is directly applicable to fraud detection where fraudulent transactions are rare. You have experience with MLflow or a similar experiment tracking platform, and you understand how to version datasets and models for reproducibility. Experience with real-time serving infrastructure — Kafka, Redis, and low-latency REST APIs — is highly desirable, as our fraud models must respond within one hundred milliseconds. Experience with explainability tools like SHAP is a strong advantage given our regulatory context. We offer a top-of-market salary, remote flexibility, GPU compute budget for research, and the opportunity to publish your work.