Native Apple product ownership
I own architecture and product decisions for software that ships on Apple platforms: framework R&D, UX constraints, App Store optimization, and distribution.
Two flagship lines on the same stack: CoreRecord (macOS Machine Learning and Private Cloud Compute for meeting intelligence) and Songen (iOS Core Audio for a commercial generative creator tool).
Project: macOS Machine Learning (CoreRecord)
CoreRecord is a macOS meeting synthesizer with local AI bounded by macOS memory and privacy limits. On-device inference uses Apple's Foundation Models; when the context window needs more, stateless Private Cloud Compute takes the request. Raw audio and transcripts do not pass through third-party APIs.
CoreRecord records system and microphone audio, transcribes on the Neural Engine within thermal and memory limits, and sends sanitized text to Private Cloud Compute for executive summaries.
Project: iOS Core Audio & Generative (Songen)
Songen runs DAW-style programmatic synthesis on iPhone and Mac inside battery and latency limits. Generative workloads stay on Apple silicon.
I tuned Core Audio pipelines on Songen and Vibin for on-device generation. The product reached six-figure revenue as a bootstrapped App Store business while holding to iOS latency budgets.
Systems Engineering & Distribution
I work on Apple platforms only: on-device model optimization, release engineering, and App Store distribution.
Background
Before on-device machine learning, I logged 1,700+ hours of QA for enterprise tech clients and produced commercial audio catalogs that earned gold record status. That testing and audio work still shapes how I ship native apps.
Connect & Explore
- Ship Log: Ship log archive
- CoreRecord: macOS ML & Private Cloud Compute
- Songen: iOS Core Audio & generative
- Full portfolio: App catalog
- LinkedIn: Professional profile
