I build and ship product in healthcare, insurance, and banking.

Problem

Providers were abandoning Medicaid enrollment partway through. A long, confusing form with weak validation caused a lot of rework, and every field carried CMS rules.

What I did

Looked at where people dropped off, traced it to two specific points in the form, and led a redesign of the workflow and validation that kept the compliance rules but removed the friction.

Result

Processing time dropped 25% in Q2–Q3 2025, on top of a 35% reduction from an earlier round. Fewer resubmissions and cleaner throughput for the state programs.

What I'd track next

Completion rate by state program, and how often applications come back for rework after submission.

Problem

Claims were getting stuck in the application, and lead data was scattered across systems. Conversion leaked, and no one had a full picture of the customer.

What I did

Used call center data to find exactly where claims stalled, and prioritized the fixes that cleared the biggest bottlenecks. Separately, led a CRM upgrade and data migration that pulled the lead data into one view.

Result

Throughput rose 40% in the first quarter after the changes. Lead conversion went from a 30% baseline to 50% once the lead data was consolidated and hand-offs were streamlined.

What I'd track next

Whether the conversion jump holds past the first quarter, and data quality after the migration.

Problem

About 65% of the call center's volume was the same set of P&C and commercial-lines questions. Customers waited on hold for answers software could give, and staffing all those calls was expensive.

What I did

Built a conversational assistant to handle those questions. Instead of trying to cover everything at launch, I started with the 500+ most common ones so it was useful from day one.

Result

The assistant handled most of the repetitive calls on its own. After launch I kept reading the feedback and reprioritizing, and satisfaction passed 80% within six months.

What I'd track next

How often it resolves a question without handing off, and the point where it should pass the customer to a person before they get frustrated.

Problem

Password resets were a large share of contact-center volume for a mobile banking app. They cost money to handle and left locked-out customers waiting.

What I did

Worked on the requirements for a self-service reset using biometric and facial-recognition sign-in, with competitive research to match what customers already expected from other banking apps. This was a supporting role on a larger team.

Result

The release cut contact-center operating cost about 18% in the first quarter, with faster response times and a smoother experience in the app.

What I'd track next

How many customers actually use self-service, and how often sign-in fails and sends them back to a call.