How fraud actually gets caught.
Engineering deep-dives on the systems behind Finhaq, and practical compliance writing for financial institutions. No filler.
Inside Finhaq: the engineering
Four posts from the team that built the platform: the verdict pipeline, fund tracing, account takeover, and behavioral trust scores.
Follow the money: how we built real-time fund tracing
Transactions are edges, not rows. The graph model, bounded traversal, and mule signatures behind tracing stolen funds in 380ms, and freezing them at the destination.
Anatomy of a 142ms verdict: six engines, one explainable decision
How every transaction is screened by six engines inside a 200ms budget, why the engines fail open instead of down, and how every verdict stays explainable rule by rule.
Catching account takeover at the application layer
Firewalls watch the network; attackers log in. The signal taxonomy of devices, IPs, sessions, and behavior that catches account takeover inside the application, before money moves.
When behavior breaks: trust scores that freeze accounts themselves
Per-account behavioral baselines, the trust score math, and the auto-interdiction threshold that freezes an account below score 20, with humans in the loop.
Compliance, in practice
How to file a Suspicious Transaction Report (STR) with the NFIU
The 24-hour clock, goAML registration, and the narrative that decides whether your filing survives review. A practical walkthrough for Nigerian banks, MFBs, and fintechs.
Score your compliance readiness
Eight questions, three minutes, instant score. See where your reporting process would break under examination.