DORVELON | AGENTIC AI & MODEL ASSURANCE
Dorvelon measures what AI and models expose an institution to — algorithmic fairness, disparate impact, re-identification, and whether a model performed as represented — and documents the method behind the number. Led by senior principals: a former Chief Model Risk Officer and doctorate-credentialed practitioners who worked under continuous federal examination.
Book a Consultation ↗WHY DORVELON
01
Model validation and audit performed inside the GSEs and Tier-1 U.S. financial institutions under continuous federal examination — approximately 200 models across credit, allowance, fair value, servicing, and AI/ML applications.
02
Selection rate and adverse impact analysis, less discriminatory alternative search, statistical disclosure risk, outcome and performance testing. Every finding carries the method, the population, and the effect size behind it.
03
Doctorates in applied mathematics and economics, graduate statistics, and eleven years auditing machine-learning and credit models — with the method examined continuously by a federal regulator, not asserted.
04
Retained directly, never through a vendor list. Scoping, analysis, and sign-off are done by the principal who performed the work.
Engagements
01
Consulting and testifying expert work on algorithmic fairness and less discriminatory alternatives, disparate impact, re-identification risk, and whether a model performed as represented — with the method documented to survive challenge.
Learn More →02
Model governance build-out for lenders and seller-servicers without an internal model risk function — inventory, tiering, validation standards, and the control and governance requirements they are measured against.
Learn More →03
In-house training for audit and risk functions on how to audit an agentic AI system or a model — delivered to a named team, against their own model population.
Learn More →Framework Coverage