On-device AI / Product engineering

Building elegant projects with on-device intelligence.

I help indie developers and small product teams ship privacy-first iOS AI products. You will find clear implementation paths, App Review strategy, and compliant monetization patterns grounded in real shipped work.

Developer Guidance

What you'll find here

Actionable field notes for indie developers and small product teams shipping privacy-first AI products to the App Store.

On-Device InspirationNew use-case directions and UX patterns designed for local model strengths.
Practical ImplementationStep-by-step technical playbooks for landing edge models in production.
App DevelopmentNative iOS engineering lessons from real consumer product delivery.
Store & MonetizationCompliant product framing and monetization examples that pass App Review.
Privacy-First FeaturesProduct opportunities uniquely enabled by on-device processing and local intelligence.
Knowledge Base
11
Long-form engineering notes Architecture trade-offs, runtime decisions, and practical delivery patterns.
Review Intelligence
6
App Review field notes Real rejection-to-fix loops from shipped iOS AI products.
Product Evidence
6
Live App Store product cases Each case mapped to stack, model path, and implementation constraints.
Technical Thesis
1
On-device AI as product engineering GGUF + llama.cpp with Apple Foundation Models in production contexts.
Featured Writing

Notes on models, architecture, and elegant implementation.

These are working notes on how to build better software with clearer technical judgment.

See all writing

How CompaniPet Got Caught Between AI Companion UX And App Review Safety Rules

The hardest App Review rejection for CompaniPet: AI Pet Warm Chat was not a crash, a broken purchase, or a missing screenshot. It was a product framing problem. Compan...

App Store ReviewCompaniPetSafety
Read article

The Subscription Rejection In Should I Decide Was Really A Value Explanation Problem

The first subscription rejection for Should I Decide: No Dilemma looked like a StoreKit problem. It was not. StoreKit worked. The product identifiers were correct. The...

App Store ReviewSubscriptionsStoreKit
Read article

AURA UP And The Metadata Rejection That Forced Me To Make The Screenshots More Honest

The rejection for AURA UP: Easy Be More Pretty was uncomfortable because the app worked. The local image analysis pipeline ran. The onboarding loaded. The screenshot f...

App Store ReviewMetadataAURA UP
Read article

The App Review Rejection That Made Easy PDF Pro's Privacy Story Much Sharper

I used to think privacy was one of the easier parts of submitting Easy PDF Pro: On Device AI to the App Store. The whole product was built around a simple promise: let...

App Store ReviewPrivacyAI
Read article
App Review

A field-notes column from real iOS submissions.

What was rejected, what changed, and what the review process made clearer.

Open column

How CompaniPet Got Caught Between AI Companion UX And App Review Safety Rules

The hardest App Review rejection for CompaniPet: AI Pet Warm Chat was not a crash, a broken purchase, or a missing screenshot. It was a product framing problem. Compan...

App Store ReviewCompaniPetSafety
Read article

The Subscription Rejection In Should I Decide Was Really A Value Explanation Problem

The first subscription rejection for Should I Decide: No Dilemma looked like a StoreKit problem. It was not. StoreKit worked. The product identifiers were correct. The...

App Store ReviewSubscriptionsStoreKit
Read article