About

1) Summary We’re hiring a product‑minded mobile developer (or small team) to design and build a polished ChatGPT‑wrapper app. Users can upload an image or paste text, set basic filters/preferences, and receive ranked recommendations with short “why this” explanations. The app stores a lightweight history & feedback loop so results improve over time. No separate OCR stack required — use the model’s vision capability to read and reason over images. We’ll provide product specifics during kickoff. 2) Responsibilities Product & Design Translate a short brief into flows, wireframes, and Figma. Create a clean, modern, approachable design system (light/dark). Mobile App (iOS & Android) Build with React Native (Expo preferred) or Flutter. Implement camera/upload, image handling, and robust error states. AI Integration Integrate OpenAI (GPT‑4o/4.1) or equivalent for image + text understanding and recommendation reasoning. Enforce structured JSON outputs and resilient prompt handling (timeouts/retries). Core Features User profile & preferences (simple schema). Recommendation flow (inputs → ranked list + “why”). Feedback loop (like / not for me) and history. Filters (e.g., budget/occasion) and shareable result card. Platform & Operations Auth (Email/Apple/Google). Subscription paywall (RevenueCat or native IAP). Analytics & crash reporting (PostHog/Amplitude + Sentry/Crashlytics). CI/CD, environment management, and basic admin utilities. App Store & Play Store submission support. Docs & Handover README, setup, architecture diagram, prompt docs, test plan. 3) MVP Scope (Domain‑Agnostic) Inputs: Image upload (or camera) and/or pasted text. Reasoning: Model interprets content and produces a ranked, personalized list. Explanations: Short rationale for each recommendation. Profile: Lightweight preferences; improves via feedback. History: View past sessions and outcomes. Paywall: Usage limits on free tier; subscriptions unlock more. Polish: Empty/skeleton states, accessibility basics, shareable card. 4) Tech Stack (Options + Pick) Option A — Recommended: React Native (Expo) + Supabase (auth/DB/storage) + OpenAI GPT‑4o/4.1 + RevenueCat + PostHog/Sentry. Why: Fast cross‑platform velocity, OTA updates, low ops, strong ecosystem. Option B: Flutter + Firebase + OpenAI + RevenueCat. Why: Great UI fidelity; similar capabilities; pick if you’re stronger in Flutter. Option C: Native iOS/Android + custom backend. Why: Max control; longer timeline/cost — only if you can justify clear wins. Please propose your stack and why. 5) Deliverables & Milestones M1 – Design (Week 1–3): Flows, Figma, design tokens, clickable prototype. M2 – Foundations (Week 2–4): Project scaffold, auth, nav, backend setup, analytics, error system. M3 – Vision + Reasoning (Week 3–6): Image/text input → structured JSON → ranked recommendations with explanations. M4 – Profile/History/Paywall (Week 5–8): Preferences, feedback loop, usage limits, subscriptions. M5 – Polish & QA (Week 7–10): Performance, accessibility pass, test plan, bug fixes. M6 – Submission (Week 8–12): App Store/Play listings, build & submit, handover. API Keys: We can provide OpenAI keys; if you prefer to develop using your own keys initially, that’s fine. We’ll switch to our keys before release. 6) Acceptance Criteria (Excerpt) Image understanding works on clear, standard text‑based images (model vision only; no custom OCR). AI returns valid JSON alongside user‑facing copy; errors degrade gracefully. Cold start, less than 2.5s on mid‑range devices; input → recommendations less than 8s on typical LTE/Wi‑Fi. Crash‑free in 200 consecutive test sessions; memory leaks monitored. Clean repo, linters, TypeScript (if RN), unit tests for parsing/reasoning, reproducible builds. Work‑for‑hire; all IP/assets/code transferred; third‑party licenses documented.