Min-cut on a 4.49M-transfer graph to freeze mule accounts (and a max-flow bug)
JeremyCMorgan · x · 2026-10-29
Developer Omkar Rane published a pre-registered, honest-results writeup of Chokepoint, an anti-money-laundering tool built on FalkorDB:
- Problem: after a scam alert on a mule account, which few accounts do you freeze to cut money's escape routes without locking half the bank? Framed as a max-flow/min-cut problem
- Data: IBM synthetic AML transfers, 5.08M rows, 4.49M loaded in FalkorDB; each case runs the cut on a small subgraph of tens to thousands of accounts
- Honest result: the graph-based freeze list blocked roughly the same money as "freeze recent payees" but was far more precise about who was laundering—a model example of reporting a tie
- Bug found: FalkorDB's algo.maxFlow returns wrong answers in a certain case
- Engineering: node splitting converts edge cuts into account cuts; every plan double-checked, human-approved, approvals can't outlive evidence
Code, live demo, and a one-command reproduction of every number are available.
Related event: AML Project Applies Min-Cut to 4.49M Transfers, Uncovers FalkorDB Bug(2 posts)→
More from coding & agent
- Weaviate Podcast: MIT PhD's "Machine Studying" lets always-on agents prep without knowing the task — CShorten30 · 2026-10-29
- Dev rebuilds Slack as video with Cursor agents and computer use — mattyp · 2026-10-29
- Deepgram acquires Daily, spins up independent Pipecat Foundation — charles_irl · 2026-10-29
- Don't force multi-agent for enterprise: a one-night RPA script beats slow, costly agents — sujingshen · 2026-10-29
- After 30 approval prompts, 'yes' to everything sent a broken release to his whole list — latent_lurker8587 · 2026-10-29
- Nat Eliason builds a personal AI writing system in one Claude Code session — nateliason · 2026-10-29