FitNest

Overview

Fitnest (Holistic Health Buddy) is a production-ready health and wellness platform that consolidates fitness tools, mental health resources, and AI-powered nutrition planning into a single, cohesive web experience. It was built as a full-stack showcase project covering the full development lifecycle — from database schema design and authentication to public deployment.

Problem

Health and wellness tools are typically fragmented across many separate apps — one for calorie tracking, another for yoga, another for meditation. Users who want a holistic approach to well-being have no single place to manage it all. The goal was to build a unified platform that addresses physical health (calculators, nutrition), mental health (meditation, mood tracking), and active wellness (yoga library) — all behind a secure, personalized account.

Solution

The application was architected as a Next.js full-stack app, with pages-based routing for public and authenticated experiences. Supabase handles both the PostgreSQL database and authentication (email/password with confirmation flow, password reset, and session management). A custom ProtectedRoute component enforces access control across all private pages. AI-powered meal planning is served through a Next.js API route that proxies requests to OpenRouter, keeping API keys server-side and allowing model flexibility. Yoga and meditation content is stored in Supabase and served via versioned REST API routes (/api/v1/), seeded from JSON files using a custom migration script.

Tech Stack

• Next.js 15 + TypeScript — Full-stack framework with API routes, SSR, and strict type safety across the entire codebase.
• Supabase (PostgreSQL + Auth) — Serverless database and auth with row-level security, chosen for its generous free tier and built-in email auth workflows.
• Tailwind CSS + Radix UI — Utility-first styling paired with accessible, unstyled primitives for a consistent, WCAG-friendly UI.
• TanStack React Query — Server state management with caching and background refetching for data-heavy pages like the yoga and meditation libraries.
• React Hook Form + Zod — Declarative, schema-validated forms with minimal re-renders and strong runtime type guarantees.
• Framer Motion + Lottie React — Smooth page transitions and micro-interactions without sacrificing performance.
• Recharts — Lightweight charting for BMI/BMR and nutrition data visualizations.
• OpenRouter API — Abstraction layer over multiple AI models for the meal planner, enabling cost-flexible AI responses without vendor lock-in.
• Vercel — Zero-config deployment with automatic preview deployments per branch.

Challenges

1. Secure authentication end-to-end Implementing the full auth lifecycle (signup, email confirmation, login, password reset, session persistence) correctly required careful coordination between Supabase's redirect URLs, Next.js API callbacks, and client-side session hydration. The auth callback at /auth/callback handles token exchange and redirects users to the correct post-auth destination.

2. Strong password validation Beyond basic length requirements, the app needed to detect common or easily guessable passwords without a full HIBP API call on every keystroke. A client-side utility (src/utils/passwordValidation.ts) runs complexity checks and a local common-password list, with the architecture designed to support a server-side HIBP check as a future enhancement.

3. AI meal planner on a serverless budget Streaming AI responses from a third-party provider through a Next.js serverless function required managing timeout limits on Vercel and structuring the prompt to return structured JSON (diet type, cuisine preference, macros, allergen exclusions) rather than free text — making the response directly renderable without parsing overhead.

4. Data seeding and migration Yoga poses, meditation practices, and mood data needed to be loaded into Supabase from static JSON files without manual SQL. A custom lib/migrate.ts script handles upserts with conflict resolution, enabling safe re-runs during development and schema changes.

Results / Impact

• Deployed to production on Vercel at full feature parity.
• Complete authentication system covering signup, login, email confirmation, password reset, and protected routing — no third-party auth UI dependencies.
• AI meal planner supports 4 diet types, cuisine preferences, and allergy management, returning structured weekly plans via a single API call.
• Yoga library with categorized poses and difficulty levels, and a meditation library with guided practices — both served from a versioned REST API backed by Supabase.
• Codebase follows strict TypeScript throughout, with Zod schema validation at all form and API boundaries.