Local verification
Bionic Lab 01
AML Review Engine Lab
A practical blockchain AML triage demo. The useful part is not only the score — it is the control layer around the score: rules, anomaly signals, evidence notes, triage labels, and review limits.
Visual proof
Control map
The lab is framed around a reviewable trail, not a black-box risk score.
Detection claim in repo
94.9% overall detection rate
Rules documented
28 rules
Analyst queue reduction
63% triage reduction
Control layer
What this lab proves
The lab shows how a crypto-risk workflow can move from raw activity to a reviewable record. It separates detection from decision, which is the core Bionic Banker principle: software can surface risk, but a human-controlled process owns the final interpretation and next action.
Review flow
From wallet activity to review queue
1. Input
Wallet or transaction activity enters the review lane.
2. Rules
Known patterns flag threshold, velocity, sanctions, mixer, bridge, phishing, or structuring signals.
3. Anomaly layer
Isolation Forest-style anomaly detection adds a second view for unusual graph behavior.
4. Triage
Signals become priority labels so human review focuses on the highest-value cases.
5. Evidence note
The output must explain what was seen, what sources support it, and what remains uncertain.
6. Human boundary
The system informs review. It does not approve, file, freeze, trade, or accuse.
Verification
Latest local check
The engine was verified locally before this public demo was published.
git clone https://github.com/hash02/aml-detection-engine
cd aml-detection-engine
python3 -m pytest -q
.............................................. [100%] The local repository is currently behind its GitHub remote and has uncommitted changes, so the next engineering step is to reconcile the repo before publishing a packaged release.
Limits
What this lab does not do
- It does not provide legal, regulatory, tax, compliance, trading, or investment advice.
- It does not identify a private actor behind an address without off-chain evidence.
- It does not freeze funds, file reports, approve KYC, or take enforcement action.
- It does not claim that every suspicious pattern is illicit activity.
Related
Wallet risk notes
Read the public wallet-risk explanation and how risk rows should be interpreted.
Open wallet risk →Next
Package the demo
The next step is a sanitized demo bundle: sample input, sample output, risk memo, and visual control map.
Back to Labs →