At Gamaka Intellectual Machines, I work with medical advisors and claim reviewers to turn clinical and compliance rules into testable requirements for a medical insurance claims platform. I help define which decisions can be automated and when a case should go to a medical officer.
For the Medical Insurance Claim Validation Engine, I made cited evidence and abstention central to the reviewer workflow: findings without source quotes return to a medical officer, while clear cases can be reviewed in a dashboard with approve, query or deny actions. In internal testing, review time fell from about 15 to about 5 minutes per claim.
I also shaped a fraud and AML triage product that screens alerts before they reach an LLM pipeline, with a two-call cap per escalated alert. Its ML tier cleared about 70 percent of alerts before the LLM while keeping recall above 90 percent on high-risk cases.
