Sentinel: AI Fraud Detection & Sanctions Screening

The results

55% cost reduction
Cost-optimized decision thresholds vs. baseline

sub-50ms scoring latency
Real-time fraud scoring with on-demand SHAP explainability

97.5% sanctions precision
Screen against OFAC lists with confidence

ROC-AUC 0.8861
Exceeds the 0.85 industry benchmark for fraud detection

The problem

Payment companies face impossible tradeoffs:

Fraud is rising: Card-not-present fraud costs cross-border payment companies 1.5% of transaction volume ($1.5M lost on $100M volume)

False positives kill conversion: Legacy systems flag 10-20% of good customers

Compliance is complex: OFAC screening must happen in milliseconds

Traditional rule-based systems can't keep up with evolving fraud patterns.

The solution

Sentinel combines ML-based fraud detection with real-time sanctions screening:

Fraud detection

LightGBM gradient boosting trained on 429 features
Explainable AI (SHAP values) for regulatory compliance
Calibrated probabilities with cost-optimized thresholds

Sanctions screening

Fuzzy name matching against 39,350 OFAC entities
Handles typos, name variations, and transliterations
Sub-50ms screening with 97.5% precision

Production-ready API

FastAPI with async processing
Redis caching for sub-200ms latency
PostgreSQL audit logs for compliance

How it works

Simple process

Submit a transaction via REST API or demo UI
1. Get fraud score in under 50ms, with on-demand SHAP explainability
Accept, decline, or flag for manual review with full audit trail

Technical implementation

Built with LightGBM for fraud detection, fuzzy matching for sanctions screening, and FastAPI for real-time API performance. Includes:

Instant results: fraud risk score, SHAP explanations, sanctions match
Transparent decisions: auditable predictions for compliance teams
Production stack: Redis caching, PostgreSQL audit logs, Next.js dashboard

Try live demo | Book pilot call

Research project disclaimer

WARNING

This is a research case study demonstrating production-grade ML engineering. Models are trained on public datasets (IEEE-CIS, PaySim) for non-commercial research use only.

For production deployments: We train custom models on your transaction data, achieving 20-30% higher accuracy by learning your specific fraud patterns.

Pilot Sentinel with your transaction data

We're offering free pilots to Series B-D payments companies.

What you get

Custom fraud detection model trained on your transaction data
2-3 week side-by-side comparison with your current system
Detailed analysis: accuracy improvements, false positive reduction, cost savings

What we need

Access to 3-6 months of historical transaction data (anonymized customer data accepted)
Weekly check-ins to review results and gather feedback
A testimonial if we reduce false positives by 20%+ or improve fraud detection by 15%+

No cost. No long-term commitment. If we don't hit the improvement threshold, you've lost nothing. If we do, you get early access to production-ready fraud detection built for your specific patterns.

Spots available: 5 companies per month

Book pilot kickoff call

Try it yourself

Try live demo — Test with sample transactions

View source code — Apache 2.0 licensed

About Devbrew

We build production AI infrastructure and agents for payments companies. Sentinel demonstrates our technical depth in ML engineering, real-time systems, and regulatory compliance.

Questions? Contact us | View repo: GitHub | LinkedIn