Phemilo

Why I built it

Phemilo started after another strength-training app lost weeks of my recorded workouts. The missing history was more than an inconvenience: progressive training depends on being able to trust what happened before.

I decided to build the product I wanted to use myself. What began as a personal solution became a native iOS and watchOS application that I designed, built and released independently.

Decisions I made

• I chose native SwiftUI for iPhone and Apple Watch so logging a set could feel at home on both devices.
• I treated training history as core product data rather than a disposable activity feed.
• I kept the backend, subscriptions and entitlements within an architecture I could operate as a solo developer.
• I built coaching around workout-history analysis, plan generation and adaptation, exercise retrieval and progress-based recommendations.
• I added PII redaction and Zero Data Retention safeguards around the model integration.
• I use traces, datasets, reference answers, automated evaluators and human review to compare changes before they reach users.

What I own

I work across the entire product: deciding scope, shaping the interaction, writing the native apps, evolving the data model, integrating AI workflows, managing subscriptions and shipping releases.

Phemilo is still active product work. The interesting part is not that every decision was perfect; it is that I can follow those decisions all the way from a user problem to software running in production.