Product manager · builder
Three consumer apps, live and in use — an AI training coach, a social planning tool, and a baby sleep scheduler. Each one takes a plan people care about and adapts it to the day they're actually having. I do the product thinking, the design, and the code.
All three are shipped, running products — not mockups. Every case study links to the live thing.
AI training coach — flagship case study
Reads your sleep, recovery, race calendar, and how you say you feel, then prescribes the one workout your body is ready for — with the reasoning stated in your own numbers. An eval suite gates every coach change; a deterministic engine backstops the AI.
Make plans that actually happen
Group threads die and another month goes by. HangEasy gives the organizer a two-minute flow and guests a zero-friction, no-account RSVP — because the plan only happens if responding is easier than ignoring.
Nap schedules without the math
Log a nap as it happens and the rest of the baby's day recalculates — age-aware wake windows that learn your baby's actual rhythm. Free forever, honestly sourced, built for my own daughter. 5.0 on the App Store.
Every product above is one core loop, polished until it's automatic: check in → get a plan → live your day → the plan adapts. Features only ship when they make the loop tighter.
FitScript's coach ships behind an eval suite that encodes the safety and quality bar as executable scenarios — it has already caught real gaps before users could. "Accurate and safe" is a test result, not a vibe.
A rough night with a newborn is real training load. A guest who won't download an app is a design constraint. How people feel and behave is first-class data, and the products treat it that way — without guilt.
I run these products end to end — research, product decisions, design, code, and iteration in production. Want to talk? emilydenis43@gmail.com