How laurien03/enterprise-aml-pipeline works
AML-Enterprise is a transaction-monitoring system that prepares transaction data, enriches it with anomaly and network features, trains graph and tabular models, and fuses risk scores into flagged cases. The primary backend workflow starts in main.py and is orchestrated by pipeline_engine.py; a separate Streamlit app presents stage visualizations. The graph reflects the sampled pipeline and UI wiring. The README describes the product goals but not additional runtime connections, so unsampled capabilities and relationships are not inferred.
Main entry points: pipeline_engine.py AML pipeline · app.py Streamlit visualizer · main.py Backend entry point · data_manager.py Data preparation
Components 16
Pipeline orchestration
- Backend entry point
main.py - AML pipeline
pipeline_engine.py
Transaction preparation
- Data preparation
data_manager.py
Detection and graph models
- Anomaly scoring
anomaly_detector.py - Transaction graph
graph_analyzer.py - GraphSAGE model
fraud_classifier.py - XGBoost model
fraud_classifier.py
Risk review and UI
- Risk fusion
compliance_reporter.py - Investigation board
compliance_reporter.py - Streamlit visualizer
app.py - Ingestion stage view
data_ingestion_flow.html - GraphSAGE stage view
graphsage_neural_engine.html - XGBoost stage view
xgboost_tabular_engine.html - Risk radar view
compliance_risk_radar.html
Other components
- Operator
- Transaction CSV
Connections 17
- Operator → Backend entry point: starts
main.py - Backend entry point → AML pipeline: runs
main.py - Operator → Streamlit visualizer: opens
- Transaction CSV → Data preparation: reads
data_manager.py - AML pipeline → Data preparation: prepares data
pipeline_engine.py - AML pipeline → Anomaly scoring: scores splits
pipeline_engine.py - AML pipeline → Transaction graph: builds graph
pipeline_engine.py - AML pipeline → Transaction graph: extracts features
pipeline_engine.py - AML pipeline → GraphSAGE model: trains and embeds
pipeline_engine.py - AML pipeline → XGBoost model: trains and evaluates
pipeline_engine.py - AML pipeline → Risk fusion: fuses scores
pipeline_engine.py - AML pipeline → Investigation board: constructs
pipeline_engine.py - AML pipeline → Backend entry point: returns results
main.py - Streamlit visualizer → Ingestion stage view: renders
app.py - Streamlit visualizer → GraphSAGE stage view: renders
app.py - Streamlit visualizer → XGBoost stage view: renders
app.py - Streamlit visualizer → Risk radar view: renders
app.py