iOS and Android Apps

We want to build 2 private apps where AI runs entirely on the user's phone — no cloud, no accounts, nothing leaves the device. We're hiring a contractor to build two apps (Local Notes and Local Money) on a shared foundation, on top of Tether's open-source QVAC SDK. We already have a clickable prototype and a detailed spec, so the vision is defined — we need someone who can build it. What you'll build: Two consumer apps on iOS and Android, sharing one core (built once, reused). On-device AI via the QVAC SDK: local LLM chat over the user's own notes/data, plus OCR and transcription. Retrieval (RAG) over local, encrypted storage. No backend, no login. A no-account onboarding where the user downloads an AI model on first run (mirroring Tether's own qv.ac / QVAC Health apps). How the engagement works — 3 milestones We pay per milestone. Milestone 1 is a small, paid proof-of-capability that comes before anything else — it's the hardest technical part, so we tackle it first. We only continue to Milestones 2 and 3 once it's done. Milestone 1 — Prove the on-device AI works (small fixed fee). Stand up a minimal Expo app that integrates the QVAC SDK and, on a real physical device, downloads a model and runs one streamed on-device LLM completion. No product features yet — just proof the foundation runs on hardware. Deliverable: the running build + a short note on the setup. Milestone 2 — Deliver Local Notes. The full first app on iOS + Android, built on the shared core: notebooks/notes, a simple block editor, and the headline feature — AI chat over your own notes, on-device. Plus the no-account model-download onboarding and encrypted local storage. Milestone 3 — Deliver Local Money. The second app on the same core: manual/receipt-based finance with on-device OCR for receipts, local categorization, and AI chat about your spending. 100% on-device (no bank aggregation in v1). Must-haves Shipped React Native + Expo apps on both stores (SDK 54+, New Architecture, TypeScript). Real experience running AI models on-device on mobile (llama.cpp / whisper.cpp / ONNX Runtime / MLC / ExecuTorch / Core ML — or the QVAC SDK itself). Comfort with native modules / Expo config plugins and custom native builds (not Expo Go). Your own physical iPhone (iOS 17+) and Android device (12+) for testing — on-device AI can't run on simulators. Nice-to-have SQLite (op-sqlite) + vector search (sqlite-vec) or embeddings/RAG on mobile. Prior local-first / privacy-focused app work.