About

About three and a half years ago, my life changed: ChatGPT arrived. At first I used it for emails and poetry, but as the models got smarter, so did what I could do with them, summarizing dense papers, explaining concepts I'd never quite grasped, turning "I don't get this" into "oh, that's actually simple." We were all warned it would hallucinate. It did. I pressed on anyway.

When the "thinking" models showed up, my software skills leveled up with them. I started writing automations left and right. Then Claude Code and Cursor arrived, and I started cooking. Not clever prompts, real workflows that ran on their own.

Today I'm the lead AI Solutions Engineer on the AI Enablement team at WebMD. I embed with teams and map how they actually work before I touch any code. My flagship project is BuildDiff, a deterministic comparison engine that checks front-end builds for every kind of change: text, styling, code, layout, and tracking links. It runs before anything goes live in a regulated medical content environment. I layered an AI triage step on top with the Claude API to summarize what changed in plain language, but the deterministic core is the part that has to be right every time. Building it taught me as much about hosting on legacy infrastructure and working around privacy constraints as it did about model design.

Along the way, I also led WebMD's Perplexity Pro rollout and became a Perplexity Business Fellow. On my previous team, the knowledge base and automations I built became the groundwork for what's now a dedicated Digital Assistant Twin project. This past year I also ran AI Literacy Camp, demos and prompting workshops that reached 700+ people across my entire organization, plus playbooks people could actually use with Gemini and Google's AI tools. I bring the same instinct to local AI meetups in NYC, where I show up and sometimes speak: if something's working for me, I want it in other people's hands too.

I lead with implementing, shipping working systems into production, and treat teaching and rebuilding legacy processes as the supporting moves that make it actually stick. Abstract AI strategy that never ships doesn't interest me. What does: understanding the humans and systems already in place, then building the thing people will actually use.