InariSense

InariSense is a cross-platform gardening app that goes beyond a static planting calendar. It combines location-aware planting guidance, camera-based plant identification, and weather-adjusted care reminders, with an explicit focus on explaining why a recommendation was made and how confident the app actually is in it. The backend is built with FastAPI and SQLAlchemy, using Alembic for versioned schema migrations, with a modular architecture organized by feature domain rather than by technical layer. External integrations (geocoding, USDA hardiness zone resolution, plant identification via the Pl@ntNet API) are built behind a provider-adapter pattern, so the underlying vendor can be swapped without touching business logic. The recommendation engine is deliberately rule-based and structured-data-first: every planting suggestion carries a confidence level and a note about its data source, and when real data isn't available (for example, there's no free, keyless API for frost dates), the app falls back to a clearly labeled estimate rather than presenting a guess as fact. The mobile app is built in Flutter with Riverpod for state management, backed by a custom design system (color tokens, typography scale, and a shared status/confidence visual language used consistently across every screen). A session bootstrap creates a persistent guest identity on first launch, so the app is usable immediately without a login flow. The project follows a disciplined GitFlow workflow with small, single-purpose pull requests, a CI pipeline that runs linting and automated tests (including a test that runs real database migrations, not just mocks) on every PR, and a Kanban board tracking every feature against the original product requirements and user journeys.