Blog
Technical writing on federated learning security, adversarial ML, and fraud detection for financial systems
Lessons from 3 Years of Production Fraud Detection
Real-world lessons from managing fraud detection systems at one of Indonesia's largest banks — covering false positive management, SLA maintenance, incident response, and the gap between academic ML and production reality.
Building a Research Portfolio in Federated Learning Security
Lessons from building a comprehensive federated learning security research suite — covering attacks, defenses, and privacy techniques across 165,000+ lines of research-grade code.
Why Banking Fraud Detection Needs Federated Learning
From 3+ years managing fraud detection at one of Southeast Asia's largest banks: why single-institution models fail, and how federated learning enables collaborative defense without exposing customer data.
SignGuard: Designing Cryptographic Defenses for Federated Learning
Deep dive into SignGuard's architecture — combining ECDSA digital signatures, multi-factor anomaly detection, and time-decay reputation scoring to defend federated learning against poisoning attacks.
Understanding Byzantine Attacks in Federated Learning
A comprehensive guide to Byzantine attacks in federated learning systems—covering data poisoning, model poisoning, backdoor attacks, and gradient leakage with defense strategies.