I calibrate alert thresholds across ML, rule-based and GenAI models, cutting false positives.
At BNP Paribas I built a model from about 15 million hits that cut false positives 43.6% with no theoretical risk.
I joined a group of mentors teaching it online through the DPhi community in spring 2020.
In a Medium series I answer data scientists' questions about build systems and dependency management.
Complidata trusted me with a student role in 2017. I ended up building teams that shipped production-ready compliance AI.
We used LIME and SHAP to explain black-box models, which helped win two top Belgian banks as clients.
My KU Leuven thesis, done with Complidata, used interpretable models built for AML investigators.
Model Governance and Analytics Manager at Interactive Brokers
I build and govern machine learning models for banks' financial-crime compliance.
I calibrate alert thresholds across ML, rule-based and GenAI models, cutting false positives.
At BNP Paribas I built a model from about 15 million hits that cut false positives 43.6% with no theoretical risk.
I joined a group of mentors teaching it online through the DPhi community in spring 2020.
In a Medium series I answer data scientists' questions about build systems and dependency management.
Complidata trusted me with a student role in 2017. I ended up building teams that shipped production-ready compliance AI.
We used LIME and SHAP to explain black-box models, which helped win two top Belgian banks as clients.
My KU Leuven thesis, done with Complidata, used interpretable models built for AML investigators.
Model Governance and Analytics Manager at Interactive Brokers
I build and govern machine learning models for banks' financial-crime compliance.
I calibrate alert thresholds across ML, rule-based and GenAI models, cutting false positives.
At BNP Paribas I built a model from about 15 million hits that cut false positives 43.6% with no theoretical risk.
I joined a group of mentors teaching it online through the DPhi community in spring 2020.
In a Medium series I answer data scientists' questions about build systems and dependency management.
Complidata trusted me with a student role in 2017. I ended up building teams that shipped production-ready compliance AI.
We used LIME and SHAP to explain black-box models, which helped win two top Belgian banks as clients.
My KU Leuven thesis, done with Complidata, used interpretable models built for AML investigators.
By using this service, you agree to the Terms of Service and Privacy Policy.