DAFU

Enterprise Fraud Detection

Advanced ML-based fraud detection with sub-50ms latency, 99.9% uptime, and 10K+ TPS.

Focus Sectors

• Financial Services — banks, issuers, acquirers
• E‑commerce — marketplaces, retail, subscription
• Payments — card-not-present, wallets, PSPs
• Fintech — neobanks, lending, crypto on/off-ramps
• Travel & Ticketing — OTA, airlines, events

Key Metrics

• <50ms API latency
• 99.9% uptime
• 10K+ TPS throughput
• 95%+ detection accuracy

Last updated: October 10, 2025

Core Capabilities

• Isolation Forest, LSTM & GRU models
• Real-time stream and batch processing
• Model persistence & comparison
• Kubernetes-ready infrastructure

Upcoming Features

• Real-time API — Sub-50ms fraud scoring
• Enterprise Security — OAuth2, JWT, RBAC
• Scalable Architecture — Kubernetes auto-scaling
• Advanced Monitoring — Prometheus, Grafana, Jaeger
• High-throughput Processing — 10,000+ TPS optimization

Quick Start

git clone https://github.com/MasterFabric/dafu.git
cd dafu
chmod +x dafu
./dafu
# dafu> fraud-detection

Project Authors

• Gurkan Fikret Gunak
  Full-stack developer focused on modern web, mobile, and cloud architectures. Passionate about Next.js, TypeScript, Flutter, and NestJS.
  github · linkedin
• M. Furkan Cankaya
  Data and ML engineer working on anomaly detection, OCR and time-series systems.
  github · linkedin

Authors Tickets

Read our open letter and see how you can contribute.

View tickets →