Fraud Detection
Serasa Experian needed more scalable GPU capacity for fraud-prevention AI so it could train modern machine learning models faster, increase the number of experiments it ran each day, and improve the response time of authentication and document-analysis products used in billions of commercial transactions.
Reported impact
Serasa Experian dramatically expanded daily fraud-model training throughput and improved antifraud product performance, enabling faster document-query responses and sub-second ID-photo analysis while supporting fraud controls across billions of annual commercial transactions.
Serasa Experian expanded the fraud-prevention side of its multicloud stack onto Oracle Cloud Infrastructure so data scientists could access the latest NVIDIA GPU resources without the shortages it faced elsewhere. OCI bare metal instances, virtual machines, and OCI Data Science provide the compute and model-development environment for hyperparameter search and neural-network training across biometric and document-analysis workloads. The company uses that environment to run more than 1,000 training sessions per day, select the best fraud models, and then feed those models into authentication products that analyze identification photos and other risk signals in near real time. By concentrating model training on OCI's GPU clusters, Serasa Experian improved both experimentation throughput and the user-facing speed of antifraud services while keeping the deployment aligned with its broader multicloud operating model.
Biometric and document-analysis training data is processed on OCI GPU clusters through OCI Data Science, where Serasa Experian runs large hyperparameter searches and daily model training; the resulting models are then used by authentication and fraud-prevention products to score photos and queries in near real time.
Adopting OCI GPUs also led to performance gains ... reducing query response times by 90% compared to the previous environment.
The company’s fraud prevention unit taps biometrics and document analysis, combined with advanced AI techniques.
the company performed over 1,000 training sessions per day to obtain the best model
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
In-house Sample Sep 24, 2026 Mastercard Fraudsters are using AI to find vulnerabilities, adapt tactics, and scale attacks across more payment channels, while traditional transaction-by-transaction controls struggle to catch coordinated fraud and authorized… Last year, scam losses globally were estimated at $442 billion 
DXC Sample Jun 29, 2026 NatWest Group NatWest Group needed to comply with the UK's image-exchange check-clearing legislation while still processing very high daily check volumes quickly, accurately, and cost-effectively, and it also needed stronger fraud… 50 percent reduction in check-clearing costs automated fraud detection software · desktop scanners · new software platform 
NVIDIA Sample Jun 2, 2026 Revolut Revolut needed to replace siloed fraud, credit, marketing, and recommendation models that depended on months of feature engineering with a single behavioral foundation model that could learn from raw financial event… 16.1% improvement in credit risk prediction performance from NVIDIA Llama-Nemotron-Embed-1B-v2 NVIDIA H100 GPUs · NVIDIA Llama-Nemotron-Embed-1B-v2 · Nebius AI Cloud 
