ABOUT ACORU:
Acoru redefines fraud prevention by acting BEFORE money moves .
We stand at the forefront of fighting fraud, protecting businesses and users with cutting-edge technology and smart solutions. Through continuous account analysis and classification, we identify risk using pre-fraud signals across all channels, before any transaction takes place. Our privacy-compliant intelligence network expands visibility beyond systems, empowering real-time, proactive defense.
Launched in 2024, headquartered in Madrid with colleagues working throughout Europe, LATAM, and the US.
THE ROLE:
As a Data Scientist – Fraud Detection at Acoru, you'll play a pivotal role in designing, building, and scaling the analytical and machine learning capabilities that power our proactive fraud prevention platform.
This is a high-impact, hands-on position for someone who thrives in complex, data-driven environments, enjoys deep investigation, and can translate technical breakthroughs into product value for our customers.
You'll partner closely with Product, Engineering, and Customer teams to develop models and signals that detect fraud BEFORE transactions happen, leveraging high-volume, real-time account and movement data.
WHAT YOU'LL DO:
Customer Data Analysis & Signal Generation
* Analyze customer datasets to understand behavior, risk patterns, and platform gaps
* Generate relevant datasets for modeling and experimentation
* Clean, validate, and enrich data to production-grade quality
* Perform feature engineering, creating robust fraud-representative variables
* Define and iterate on pre-fraud signals that generalize across channels and customers
Graph & Advanced ML Research
* Apply graph analysis to detect suspicious communities and mule networks
* Research and prototype advanced fraud detection methods
* Embeddings for movements and accounts
* Transformer-based models for sequence analysis
* Isolation Forest and clustering-based anomaly models
* Evaluate fraud-detection research papers and translate into production solutions
* Explore federated learning for privacy-compliant cross-customer model training