I can assist your organization transition from traditional rules-based surveillances to features-based behavior surveillances, improving performance and reducing cost.
Companies need to protect themselves from fraud, reputational risk and the possibility of fines through illegal customer and employee actions. Traditionally, rules have been implemented for surveillances which raise alerts when rules are broken. Since alerts are not scored or ranked they are of equal value until they have been analyzed. Machine learning algorithms have been used to show improvements over traditional rules-based surveillances in identifying outlier behavior and in scoring and ranking alerts. I have spent over 5 years implementing behavior-based surveillances that are more effective than rules-based surveillances, quicker to build and reduce analysis cost. These feature-based behavior surveillances have been used as the building blocks for machine-learning enhancement and predictive analytics.
Collaboration
Building the best surveillances requires the contribution of Subject Matter Experts, Compliance Professionals, Data Scientists, Business Analysts and Developers. As a strong communicator and experienced leader, I work as a hands-on orchestrator between the various project participants. Together with Compliance Professionals and other Subject Matter Experts, we define the surveillance risk and breakdown the associated behavioral elements or red flags. Together with Data Scientists we quantify the behavioral elements (features) so when combined the scoring will determine the probability that the behavior exists. Features are modeled, calibrated and refined. When the model shows satisfactory results with sampled data a prototype is built with a larger amount of data for Compliance Professionals to review and approve. This prototype becomes the basis for the production surveillance. My role extends to mentoring others in surveillance design and supporting Compliance in explaining surveillances to senior management, internal auditors and regulators.