Inspiration
I am driven by the rising need for secure, real-time compliance in financial services while preparing for AWS certification. Traditional anti-fraud tools often generate too many false positives or leave critical loopholes. Hence I envisioned a seamless integration of AI and blockchain to enhance trust, transparency, and regulatory efficiency, ensuring that financial institutions can detect and address risks faster.
What it does
It is a compliance and risk management platform built on a permissioned blockchain and powered by AI-driven anomaly detection. It can monitor transactions in real-time, assigning each a dynamic risk score. Additionally, it flags suspicious activity automatically, reducing manual workload, and maintains an immutable audit trail, so regulators and compliance officers can easily trace and verify every decision.
How we built it
The blockchain layer used a permissioned ledger (e.g., Hyperledger Fabric) to store transactions and smart contracts for automated compliance checks. The AI/ML engine integrated algorithms like Isolation Forest and gradient boosting to detect unusual patterns in transaction behavior. Lastly, the API and dashboard use Python Flask (or FastAPI) services to communicate with the blockchain and ML models, while a web-based dashboard (React/Angular) gives compliance officers instant insights.
Challenges we ran into
The rigorous thinking results in an idea of combination between blockchain immutability with AI-based risk scoring in a unified platform.Coming up with the idea requires dedication to search from multiple source, which has been the most challenging part.
Accomplishments that we're proud of
Built a system architecture that can scale to handle a large volume of financial data and complex regulatory scenarios.
What we learned
Merging blockchain and AI requires deep understanding of both cryptographic principles and data science techniques. And even the most advanced compliance engine must be user-friendly for compliance teams and stakeholders. Therefore, continuous model retraining and feedback loops are vital for refining risk detection and reducing false positives over time.
What's next for Northen Guardian
Develop the architecture to be business-ready by collaboration with financial institutions. As well as further fine-tuning in real-world conditions. There is also a possibility to implement Graph Neural Networks (GNN) to uncover hidden relationships and complex money-laundering rings.
Log in or sign up for Devpost to join the conversation.