How solunkepranav/sentinai works
SentinAI turns transaction data into AML anomaly findings, regulatory mappings, and an evidence-linked Suspicious Activity Report (SAR). The principal flow runs from the browser through the API and orchestrator to detection, compliance analysis, and narrative generation; audit records and report export support review. The frontend also retrieves masked transactions and presents an audit trail. The README documents dashboard charts and optional local Ollama support. Sampled source confirms the main API, detection, masking, compliance, audit, and export behavior; unsampled details are represented only where documented.
Main entry points: routes.py HTTP API · orchestrator.py Pipeline Orchestrator · app.js Dashboard & Forensics · report_exporter.py Report Exporter
Components 17
Analyst Experience
- Dashboard & Forensics
frontend/app.js - Printable SAR
API Service
- HTTP API
backend/api/routes.py - Web Server
backend/main.py
SAR Analysis
- Pipeline Orchestrator
backend/agents/orchestrator.py - Anomaly Detector
backend/modules/anomaly_detector.py - Investigator Agent
backend/agents/investigator.py - Compliance Agent
backend/agents/compliance.py - AML Law Base
backend/data/aml_laws.json - SAR Narrative Writer
backend/agents/writer.py - Report Exporter
backend/modules/report_exporter.py
Privacy & Audit
- Transaction Dataset
backend/data/sample_transactions.csv - PII Masker
backend/modules/pii_masking.py - Audit Logger
backend/modules/audit_logger.py - Audit Database
backend/data/audit.db
Other components
- Compliance Analyst
- Local Ollama
Connections 22
- Compliance Analyst → Dashboard & Forensics: uses
- Dashboard & Forensics → Web Server: requests pages
- Web Server → HTTP API: routes API
- Dashboard & Forensics → HTTP API: requests analysis
- HTTP API → Pipeline Orchestrator: runs pipeline
- HTTP API → Transaction Dataset: reads data
- HTTP API → PII Masker: masks transactions
- HTTP API → Audit Logger: fetches logs
- HTTP API → Report Exporter: generates report
- Pipeline Orchestrator → Anomaly Detector: runs checks
- Anomaly Detector → Transaction Dataset: reads data
- Pipeline Orchestrator → Compliance Agent: maps findings
- Compliance Agent → AML Law Base: retrieves provisions
- Pipeline Orchestrator → SAR Narrative Writer: generates narrative
- Pipeline Orchestrator → Audit Logger: logs decisions
- Audit Logger → Audit Database: writes and reads
- HTTP API → Audit Database: reads transactions
- HTTP API → Report Exporter: passes report inputs
- Report Exporter → Printable SAR: returns HTML
- HTTP API → Dashboard & Forensics: returns results
- Dashboard & Forensics → Compliance Analyst: shows findings
- SAR Narrative Writer → Local Ollama: optional local generation