aml-analytics 1.0.0

Open-source Python toolkit for AML detection and financial crime analytics — transaction graph analysis, anomaly scoring, SAR pattern matching, and SQL helpers for BSA/FinCEN compliance.

aml-analytics

Open-source Python toolkit for AML detection and financial crime analytics — transaction graph analysis, anomaly scoring, SAR pattern matching, and SQL helpers for BSA/FinCEN compliance.

Built by a financial crimes analytics practitioner for compliance teams and data professionals working in anti-money laundering programs at financial institutions.

---

Why this exists

Smaller financial institutions often lack the analytics infrastructure to detect sophisticated money laundering patterns. This toolkit provides open, reusable building blocks — grounded in FATF typologies and FinCEN advisories — that any compliance or data team can use to strengthen their AML detection program.

---

AML typologies covered

Typology	Module	Reference
Structuring (smurfing)	aml/patterns.py	FinCEN Advisory FIN-2014-A005
Layering	aml/graph.py	FATF Typologies Report 2006
Rapid fund movement	aml/velocity.py	FATF Guidance on AML/CFT 2012
Network-based laundering	aml/graph.py	FinCEN Advisory FIN-2022-A001
Anomalous transaction behavior	aml/anomaly.py	FATF Risk-Based Approach 2014

---

Modules

Module	Description
aml/synthetic.py	Synthetic AML transaction generator with labeled scenarios
aml/graph.py	Transaction network analysis — structuring rings, layering chains, centrality scoring
aml/velocity.py	Rolling-window velocity checks for burst activity detection
aml/patterns.py	Rules engine implementing FATF and FinCEN typologies
aml/anomaly.py	Unsupervised anomaly scoring using Isolation Forest and LOF
aml/sql/	Oracle and PostgreSQL query templates for AML detection

---

Quick start

from aml.synthetic import generate_transactions
from aml.graph import build_network, detect_structuring, detect_layering

# Generate synthetic transaction data
txns = generate_transactions(n=1000)

# Build a transaction network
G = build_network(txns)

# Detect structuring rings
rings = detect_structuring(G, threshold=9000)

# Detect layering chains
chains = detect_layering(G, min_hops=3)

print(f"Structuring alerts: {len(rings)}")
print(f"Layering alerts:    {len(chains)}")

---

Installation

pip install aml-analytics

Or install from source:

git clone https://github.com/Bhavesh0205/aml-analytics.git
cd aml-analytics
pip install -e ".[dev]"

---

Demo notebooks

Notebook	Description
notebooks/01_graph_analysis.ipynb	Transaction network analysis and layering detection
notebooks/02_anomaly_detection.ipynb	Unsupervised anomaly scoring on transaction data
notebooks/03_sar_patterns.ipynb	FATF typology rules and SAR pattern matching
notebooks/04_end_to_end_pipeline.ipynb	Full AML detection pipeline from raw transactions to alerts

---

Who this is for

• Compliance analysts who want to explore analytics-driven AML detection
• Data engineers building transaction monitoring pipelines at financial institutions
• Researchers studying financial crime detection methodology
• FinTech developers building AML capabilities into financial products

---

Disclaimer

This toolkit is built entirely on synthetic data and public regulatory guidance (FATF, FinCEN). It does not contain any proprietary data, internal model logic, or confidential information from any financial institution. It is intended for educational and research purposes and does not constitute legal or compliance advice.

---

Citation

If you use this toolkit in your research or work, please cite it as:

@software{aml_analytics,
  author  = {Bhavesh Awalkar},
  title   = {aml-analytics: Open-source toolkit for AML detection and financial crime analytics},
  year    = {2024},
  url     = {https://github.com/Bhavesh0205/aml-analytics},
  license = {MIT}
}

---

Contributing

Contributions are welcome — especially new typology rules, additional SQL patterns, or improvements to the anomaly scoring module. Please read CONTRIBUTING.md before submitting a pull request.

---

License

MIT — see LICENSE for details.