Kyc aml · Production

Monzo builds fraud detection ML systems nominated for Outstanding Prevention Initiative, with feature store x3000 data ingestion gain

The problem

Multiple types of financial fraud were causing devastating harm to customers and significant financial losses for Monzo, requiring a systematic ML-driven approach to detection and prevention.

Workflow diagram · grounded in source
1
Live transaction triggers inference
trigger
“the types of problems we were solving required live inference (e.g., making a prediction on a transaction)”
2
Fraud detection model predicts
ai_action
“In early 2020, around the same time that the world was closing down due to the pandemic, we shipped our first fraud detection model. After several iterations of this model”
3
Shadow deployment validates model
validation
“Working in this space has supercharged our usage of shadow deployments”
4
Right intervention triggered
output
“accurate models that are not used to trigger the "right" types of interventions are not impactful”
5
Governance review by auditors
human_review
“we've been working on a large governance project for a new risk assessment model in the financial crime space, which involves several iterations of review. We've learned a lot about what steps we needed to take, and the level of rigour r…”
6
Batch predictions over users
ai_action
“These are jobs that run predictions over a set of eligible users on a given cadence (such as daily or weekly)”
7
Feature store supplies model features
integration
“This system now powers six critical machine learning systems across several areas of Monzo, and Charlie optimised the system to speed up its data ingestion by a factor of x3000”
Reported outcome

Monzo's fraud detection system made an enormous dent in financial crime and was nominated for Outstanding Prevention Initiative at the 2021 Tackling Economic Crime Awards; its internal feature store was optimized to x3000 faster data ingestion, powering six critical ML systems across the company.

Reported metrics
Feature store data ingestion speed improvementx3000
critical ML systems powered by feature storesix
Fraud prevention impactenormous dent into this problem
Reported stack
GCP AI PlatformAWSPython
Source
https://monzo.com/blog/2021/10/28/machine-learning-at-monzo-in-2021
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Monzo's fraud detection system made an enormous dent in financial crime and was nominated for Outstanding Prevention Initiative at the 2021 Tackling Economic Crime Awards; its internal feature store was optimized to x…

What tools did this team use?

GCP AI Platform, AWS, Python.

What results were reported?

Feature store data ingestion speed improvement: x3000; critical ML systems powered by feature store: six; Fraud prevention impact: enormous dent into this problem (source-reported, not independently verified).

How is this kyc aml AI workflow structured?

Live transaction triggers inference → Fraud detection model predicts → Shadow deployment validates model → Right intervention triggered → Governance review by auditors → Batch predictions over users → Feature store supplies model features.

Related kyc aml cases