Arabyo - AML & Data Policy Agent
Link to open source: https://github.com/mardromus/Arabyo_
Link to Live Project: https://untangental-trudi-federatively.ngrok-free.dev/clusters?role=admin
Overview
Arabyo is an AI-powered Anti-Money Laundering (AML) and Data Policy Agent designed to monitor, analyze, and enforce compliance rules on financial transaction datasets.
It combines rule based systems, machine learning intelligence, and a database driven architecture to create a scalable, auditable, and adaptive AML monitoring platform.
The system is built to detect suspicious patterns, enforce dynamic compliance policies, and provide intelligent alerts all while maintaining transparency and version control of regulatory rules.
Why We Built This
Every year financial institutions lose billions to money laundering and face huge penalties because compliance systems are outdated, opaque and mostly static. The systems that are not are usually very difficult to set up and aren't even friendly for small institiutions. Financial systems are increasingly complex and traditional AML systems often suffer from:
- Static rule sets that break often
- Poor explainability
- High false positive rates
- Lack of version control in compliance logic
What if we could automatically translate real regulatory policy documents into executable logic that learns and adapts over time. That's exactly what Arabyo does and much more :
- Combined AI + policy engine
- Created a Fusion score by combining Deterministic rule engine , ML Model(LightGBM+Isolation Forest) and Graph Analysis
- Enabled alert clustering instead of isolated flags
- Introduced ruleset version control(with before/after impact verification)
- Integrated an AI-powered chatbot for data querying and investigation support with MCP containing the whole knowledge about the database.
- Role based Access according to user permissions.
With your support, we’ll accelerate development, refine feedback loops and bring enterprise-grade compliance intelligence to banks, fintechs and regulators that would lead to lowered compliance costs and strengthening of financial systems globally.
With funding for mentorship, cloud credits and access to AML datasets to take this from a prototype to production-readiness grade.
This isn’t just another analytics tool — it’s compliance you can trust, explain & scale.
Key Highlights
- AI + Rule Hybrid System
- MCP-Based Intelligent Chatbot
- Alert Clustering
- Ruleset Version Control
- Database-Driven Architecture
- Scalable & Modular Design
- Investigation-Friendly
This build was uploaded as a hackathon project
















