All case studies
Finance
Fraud Detection AI System Custom AI Solutions
A consumer-lending fintech was losing money to fraud rings that had found the gaps in rule-based filters tuned down to keep customer complaints low.
Finance
Industry
10 tools
Stack
Client
FinEdge Capital
Industry
Finance
Service
Custom AI Solutions
Stack
XGBoost, GraphQL, Neo4j
01
Challenge
“A consumer-lending fintech was losing money to fraud rings that had found the gaps in rule-based filters tuned down to keep customer complaints low.”
02
Build
We built a streaming detection layer that scores every transaction under 200ms with a gradient-boosted ensemble plus a graph-network signal for device and identity collusion - borderline scores queue for a small fraud-ops team, high-confidence fraud is blocked at the edge.
03
Outcome
94% precision with under 1% false positives, and US$2.8M in confirmed fraud blocked in the first six months.
Deliverables
What the system does — functionality shipped.
- 94% precision on fraud blocks with under 1% false-positive rate. US$2.8M in confirmed fraud blocked in the first 6 months. Sub-200ms scoring latency across the payment flow. Fraud-ops backlog cleared 5x faster than before.
Technologies
XGBoostGraphQLNeo4jApache FlinkKafkaPythonFastAPIRedisGrafanaModel monitoring (Evidently AI)
Similar Build?
Tell us your workflow. We'll review with this example as context and suggest the best path forward.
More Work
View all
Case study © AorBorC Technologies · aorborc.com · +1 (872) 267-2672 (872-AORBORC)