Healthcare AI & Operations | Building automation infrastructure for clinical and revenue cycle workflows

I work at the intersection of AI deployment, clinical operations, and workflow automation — focused on the problems that matter most in mental healthcare: keeping providers credentialed, getting claims paid, and making sure AI-drafted responses to patients are safe before they go out.

These projects were built to demonstrate how I think about designing operational systems that scale without scaling human effort proportionally.

AI-powered QA tool for support ticket responses at a mental healthcare platform. Scores tone, clarity, and accuracy. Flags safety concerns where a response fails to handle signs of patient distress. Audits routing decisions before responses reach patients, providers, or payors.

Monitors credential expiration dates across a provider network. Generates urgency-calibrated outreach across four tiers. Runs automated verification checks on renewal submissions. Routes only genuinely complex cases — expired credentials, payor credentialing renewals, anomalous submissions — to a human.

Tracks outstanding insurance claims and automates the follow-up layer. Generates escalating payor outreach calibrated to days outstanding. Classifies denial reason codes and routes each to the correct resolution path automatically. Surfaces high-dollar escalations and complex denials to a human billing team.

Clinical operations and AI deployment in regulated healthcare environments. Columbia MPH. I think about operational infrastructure the way an engineer thinks about systems: what breaks at scale, where the human bottlenecks are, and how to design around them.

All projects built with Python and the Anthropic API.