Local inference
React Native ExecuTorch runs a small language model and MiniLM embeddings on the device. Local processing brings privacy benefits along with real latency, battery, and device-capability constraints.
Personal project · iOS beta
Remen
Remen is a personal notes app for capturing text, voice, and scans, then finding them with natural-language search. AI inference and embeddings run on the device; iCloud provides optional backup and sync.
Under the surface
The thinking behind the work
Search beyond exact words
SQLite-backed notes support keyword, semantic, and natural-language search, reducing the need to remember the exact wording of a thought.
Persistence is a separate concern
Notes live in local storage, with optional iCloud backup and sync. The model provides inference; the storage layer provides continuity.
Have a good problem in mind?
I’m interested in senior engineering roles, thoughtful teams,
and software that’s worth building.