Turning AI strategy into products, operating models, and measurable results.

A request to sell textbooks online revealed an opportunity to own the on-campus experience that larger competitors couldn't easily replicate.

I helped turn that opportunity into an e-commerce business generating approximately $1 million in its first year, then expanded it into a multi-campus platform and built the teams to support its growth.

The important decision wasn't simply to build a website. It was recognizing which part of the customer experience was worth owning.

When COVID closed school buildings, I conceived Rooms to connect individual classrooms with families.

The product opportunity was clear, but the commercial assumptions weren't. We needed to prove that school districts would pay for a product with a different buyer and budget from our flagship offering.

I held the broader launch until we had that evidence.

Rooms grew from two design partners to 436 customers in two years.

Detecting AI-generated music was a technical challenge.

Determining what to do with the results was a product challenge.

I worked on the decision layer connecting model predictions to labeling, review, monetization, and appeals, with particular attention to false positives and their consequences for artists.

I also helped bring approximately 30 AI pilots across six functions into a more coherent roadmap and shared approach.

The original request was a support bot.

I wanted to understand why customers needed support in the first place.

I analyzed redacted support email across three brands, developed a shared taxonomy, and used the findings to help turn support into a product with clear ownership, metrics, and a roadmap.

The opportunity wasn't simply faster responses.

It was improving the experiences that generated the requests.

I led discovery across 36 conversations, identifying more than 100 workflows and 400 tasks.

We evaluated 23 AI experiments and prioritized four based on value, feasibility, risk, and ownership.

One opportunity involved Finance reconciliation, where critical rules and information were scattered across contracts, spreadsheets, and institutional knowledge.

I built a system that used AI for contract interpretation, deterministic code for financial calculations, and human judgment for exceptions.

The automated run now takes less than five minutes, compared with a previous process that took roughly a week, with Finance continuing to review exceptions.

The point wasn't to automate everything.

It was to make the process faster, more reliable, and easier for Finance to control.