NutriAgent
India-first AI nutrition and wellness platform with mobile tracking, AI coaching, subscriptions, coach/admin operations, challenges, and commerce-ready workflows.
Built a multi-surface health product across mobile tracking, AI coaching, subscriptions, coach portal, admin portal, rider PWA, challenges, privacy workflows, and commerce architecture.
NutriAgent
NutriAgent is an India-first AI nutrition and wellness platform with mobile tracking, AI coaching, Pro subscriptions, coach/admin operations, fitness challenges, and commerce-ready food-ordering workflows.
It is centered on food logging, macro targets, water, fasting, progress, AI meal planning, photo food scans, TasteMatch recipes, family profiles, Wellbeing Pulse, free step challenges, and a broader operator ecosystem for coaches, admins, riders, and commerce readiness.
Main Product Surfaces
- Customer Expo React Native app with onboarding, dashboard, food/water/supplement logging, AI coach, planner, fasting, progress, family profiles, health connections, settings, privacy, subscriptions, challenges, food scan, TasteMatch, and Wellbeing Pulse.
- Vite coach portal for dashboard, clients, client detail, plans, messages, analytics, settings, and auth.
- Admin portal for users, coaches, revenue, API keys, limits, feature flags, AI/privacy controls, commerce readiness, challenges, riders, delivery tasks, activity logs, and system health.
- Mobile-first rider PWA with availability, task queue, delivery status transitions, OTP proof, location pings, and earnings.
- Landing site with India-focused positioning and store-copy assets.
- Express API, Drizzle/Postgres schema, OpenAPI/generated clients, AI core, commerce core, and shared types.
Complexity Handled
- JWT cookie auth, email OTP, Google OAuth, roles for user, coach, admin, and rider; mobile consumer auth separated from portal auth.
- RevenueCat Pro entitlement sync/webhook, store-product config, Pro paywall; paid challenge payment provider intentionally disabled except manual test mode.
- Challenge wallet ledger, payment intents, settlement calculation/approval/posting, winner/refund/platform-fee ledger entries.
- Provider-neutral AI core, cloud Gemma, legacy Anthropic fallback, local Gemma mobile runtime boundary, structured prompts for chat, planner, food scan, TasteMatch, transformation plans, and wellbeing.
- Rider task state machine, delivery location pings, challenge step proofs, notification scheduling.
- Expo notifications for meal reminders, hydration/order-style flows.
- Coach marketplace schema/routes, gym programs, gym verification, free challenges, challenge entries, health step proofs, human coach handoffs.
- Admin dashboards for users, coaches, revenue, API keys, limits, feature flags, AI privacy, riders, challenges, and commerce readiness.
- Local SQLite storage, data export/delete, cloud account deletion, raw wellbeing/commerce telemetry rejection, legal/store privacy docs.
Caveats
Paid challenge payments are intentionally disabled except manual test mode. The product should not claim active users, revenue, store approval, or production Swiggy ordering without external proof.
What This Demonstrates
NutriAgent demonstrates full-stack product ownership: mobile UX, backend APIs, schema design, auth, subscriptions, AI orchestration, privacy boundaries, admin tooling, release planning, legal/product gating, and tests.
The strongest signal is the ability to turn a consumer app into an operating system around it: consumer, coach, admin, rider, commerce, and regulated-feature planning.
Short Version
Built NutriAgent, a multi-surface AI nutrition platform with an Expo mobile app, Express/Postgres backend, admin and coach portals, rider PWA, RevenueCat subscription flow, AI meal planning/photo food scan/TasteMatch recipes, privacy-safe Wellbeing Pulse, free fitness challenges, and Swiggy/Instamart-ready commerce architecture.
Gallery