850 work hours saved per month Generative AI Workflow Automation
A state-of-the-art generative AI system that automates workflows and improves efficiency.
Read the case study Fraud detection Cryptocurrency
Detecting crypto scams with AI. An AML system for Bitcoin and Ethereum.
An established cryptocurrency company wanted to build an anti-money laundering system for Bitcoin and Ethereum. Here’s how we helped them.
98.4%
accuracy detecting illicit transactions
The opportunity
Scalable for regulators. And for billions of transactions.
As cryptocurrencies become more important, there is a growing need for anti-money laundering solutions that traditional financial institutions can use.
Our client, an established cryptocurrency firm, chose to partner with Exometrics to build an anti-money laundering system capable of predicting the risk of cryptocurrency transactions. The solution had to be:
- Functionally scalable, to handle complex requirements from regulators
- Built to scale to billions of transactions on various blockchains
Our solution
A modular system. Machine learning and rules, running on Azure.
We designed a scalable, modular system running on Microsoft Azure. Its core combines state-of-the-art machine learning with a flexible rule-based sub-system.
- Data engineering Databases to import and manage billions of Bitcoin and Ethereum transactions
- Machine learning State-of-the-art algorithms that predict the risk of each transaction
- Rule-based sub-system Flexible rules that work alongside the models
Technologies used.
The cloud platforms, graph database, languages and machine learning framework behind the system.
- Microsoft Azure
- Google Cloud
- Neo4j
- Python
- Scala
- TensorFlow
Results
Illicit transactions, detected. With state-of-the-art models.
- Developed and delivered a sophisticated cryptocurrency AML system with state-of-the-art machine learning models
- Sourced a dataset of over 1 million illicit crypto transactions from multiple sources
- Achieved 98.4% accuracy when detecting illicit transactions
98.4%
accuracy detecting illicit transactions
1M+
illicit crypto transactions in the dataset
“Their ML skills were top-notch.”
Project reviewed: ML-based anti-money laundering tool for a digital assets firm. Read the review on Clutch (opens in a new tab)
Need a similar solution?
Get in touch for a free consultation with our experts.
Other case studies. More challenges we’ve solved.
850 work hours saved per month 
1.98 mean absolute error for ages 18–24 Age Estimation with AI
A leading age verification company wanted a more accurate age estimation system.
Read the case studyCustom-built Dashboards using Neo4j
Developing dashboards for a complex Neo4j back end with Node.js.
Read the case study