Zach Heil

I’m an analytics modeler applying machine learning in insurance, focused on turning ambiguous business problems into usable, decision-oriented models. I’ve worked across model redevelopment, feature engineering, evaluation design, bias testing, and production scoring workflows.

At Sentry Insurance, I’ve owned end-to-end remodel of the Workers Compensation Fraud referral model—from variable creation and evaluation design through production scoring implementation. I doubled the overall fraud referral acceptance rate versus the previous model version, and pioneered production use of LLM-assisted variables by designing and implementing the scoring script used to create, score, and integrate them in production.

I also designed and piloted bias testing approaches, including the BIFSG transformation, creating templates, documentation, and reusable code for broader adoption. Earlier, as a sole data analyst for Property & Casualty Reserving, I automated reporting pipelines and built dashboard foundations that improved visibility and decision support.