About Me

I've spent 17+ years in product management, taking products from first idea to scale across consumer electronics, consumer sleep products, and healthcare software. Until September 2026 I was a Senior Manager of Product Management at athenahealth, where I scaled a mobile app from 21,000 to 68,000+ monthly active users, built internal tools used by 1,200+ medical coders, and led work to bring AI into product development and clinical workflows.

I'm not a software engineer. So, I use AI coding tools, like Codex and Claude Code, to build my own products and find out what the technology can and can't do in the process. My latest, Uncork, is an offline AI sommelier that runs a small language model on my mobile phone. I wrote up what worked, what failed, and where the app had to take control away from the model.

Much of my experimentation stems from curiosity. I run open models locally, for personal and sensitive tasks, especially those with high token counts. It saves a bit of runway with paid services.

I'm optimistic about what AI could do for food security, healthcare, education and climate change. I explored the education case in a design project on female literacy in India. That optimism comes with conditions. The benefits only arrive if these systems are honest, accountable and fairly deployed, which is why I've completed coursework in AI ethics and governance and write in my Notes about sycophancy, human oversight and who is accountable when AI is wrong.

I'm also not an ML engineer, and I try to be clear about where my knowledge stops. Most of what I write comes from that boundary: where AI helps, where it doesn't, and what it takes for people to trust it.

This site is where I document the experiments, the case studies and the harder questions underneath them.

Explore: Working with AI | Vibe-Coding | Papers | Notes | Social Experiments

Contact: LinkedIn

Work

Uncork: an offline AI sommelier

Personal project, 2026

A pocket wine advisor that runs a small language model on-device. I built it with AI coding tools, evaluated it across multiple eval runs, and wrote the case study: where the model should be trusted, where the system should take over, and what the evals showed. Compact card generation dropped from 48 seconds to about 7. Read the case study (work in progress).

Building with AI coding tools

Personal projects, 2025-2026

Hush (a breathing app), Hydrolog (a hydration tracker), a World Cup predictor and a PM dashboard, each built with AI coding tools as a way to learn where AI-assisted building breaks. The consistent lesson: AI amplified my product judgement, but it couldn't supply domain knowledge I didn't have. See the Vibe-Coding page.

AI-assisted product workflows

athenahealth, 2025-2026

Designed and ran workflows that connect LLMs to Jira, Confluence and Figma through MCP, spanning requirements, build and release. One example is "death by grooming", athenahealth's quarterly ritual for preparing the next quarter's roadmap, which normally takes 3-4 days of continuous meetings across the company's 200 scrum teams. Questioning the need for that, I ran an experiment with one scrum team and condensed it into a single structured three-hour call. I adapted our ambient clinical documentation solution, using Teams and Copilot to draft meeting notes covering functional items, requirements, spikes and open questions. Prompts then turned those notes into PRDs with user stories and acceptance criteria, which went into Confluence and onto Jira boards. That saved 18+ hours in total. The case studies were presented at athenahealth's public zone review and adopted by internal teams. I also led a conversational, agentic EHR concept, a 2026 hackathon finalist.

Internal tools for medical coders (revenue cycle management)

athenahealth, 2021-2023

Shipped interactive web widgets inside the revenue cycle applications that surfaced the right clinical information in context for 1,200+ medical coders. Coders completed 1.5 additional tasks per day, a 10% efficiency gain, bringing turnaround time closer to two working days.

athenaOne Mobile: from 0 to 68,000+ monthly active users

athenahealth, 2021-2026

Built the Android app from zero to 5,600+ monthly active users, 27% above the launch target. Then designed the mobile go-to-market and engagement framework that grew monthly active users from 21,000 to 68,000+ over three years, at 85% retention.

Accessibility in clinical software

athenahealth