Director
Current- Led the model validation of the sanctions filter and customer risk rating model for a large cryptocurrency exchange. The model validation included validating input data that has been transformed through various transformation logics unique in the cryptocurrency industry- Developed the approach to perform a multiyear bitcoin on-chain suspicious activity monitoring lookback for a large cryptocurrency exchange. The approach included configuring a third party blockchain analytics tool to produce suspicious activity alerts as well as performing data analytics on the resulting alerts to focus the alert review on the riskier activities. The approach and configurations received the approval from the exchange’s regulators- Led a proof of concept (POC) to evaluate the benefit of machine learning to help automate sanctions alert disposition at the first level review. The POC work included identifying the appropriate data to use in the machine learning model development, assessing the data for quality and suitability, engineering the features that can significantly contribute to the performance of the machine learning model, and managing data scientists in performing model development and evaluating the results for suitability- Led the development of CLARITY, an in house application/tool used to assist in the execution of watchlist filter effectiveness testing. Binsar designed CLARITY to automate many manual steps involved in filter testing, such as the extraction of data from government site, the creation of name variations, and the actual analysis of the test results to identify filter’s ineffectiveness in name matching
