Hi everyone,
I’m a data analyst on a financial crime team, and I’m trying to get a machine learning project started to improve how we handle AML (anti-money laundering) work. I’d love to hear from anyone who works in transaction monitoring, customer screening, or sanctions/PEP screening and has used ML in practice.
Some context: like most teams in this space, we rely mostly on rule-based systems, which generate a lot of false positives and take up a lot of analyst time on alert review. I’d like to explore whether ML can help, but I want to start with something realistic rather than overpromising.
A few questions for those with experience:
Use cases: Where have you seen ML add the most value? For example, alert prioritization/scoring, reducing false positives, anomaly detection, entity resolution or name matching in screening, or network/graph analysis to spot linked accounts?
Data and labels: How do you deal with the lack of good labeled data? SARs are rare and confirmed cases even rarer.…