Consumer AI · Solo Build

Meal Melody

A meal discovery app for 50+ global cuisines. Scan your pantry or fridge to cook from what you already have, get recipes adapted to your diet and preferences, and follow along with a multilingual AI chef that chats, listens, and reads recipes aloud while you cook.

Visit Meal Melody Back to Home
+45%
Daily active usage after beta iteration
−30%
Load time reduction
50+
Global cuisines supported
20+
Beta users in qualitative research
The Problem

Cooking from what you have shouldn't feel like work.

Most recipe apps treat dietary preferences as filters, narrowing the cuisines you see instead of personalizing them. You lose the dishes you grew up with, or the ones you'd love to try, because the app decided they don't fit you.

And you don't always know what's actually in your own kitchen. Groceries get pushed to the back of the fridge, stacked in the pantry, lost behind something else. Cooking from what you already have starts feeling like extra work to figure out, so eating out wins.

Meal Melody flips both. A pantry and fridge scanner surfaces what you actually have, including what you forgot. Recipes get adapted to your diet and preferences instead of being filtered out. And a multilingual AI chef walks you through cooking with chat, voice, and read-aloud, so you don't need clean hands to follow along.

How It Started

Built from a problem I kept hearing.

My friends and I kept having the same conversation, one that reflected a frustration I was facing myself. We were tired of eating the same meals or constantly defaulting to takeout. The problem was real and recurring, yet no one had built a meaningful solution. I decided to build one.

Features

Four parts working together.

  • Personalization
    Set your diet, allergies, and preferences once. Every recipe Meal Melody surfaces is shaped around your profile from the start, so you never have to filter through what does not apply to you.
  • AI Scanner
    Scan your fridge or pantry, and Meal Melody recognizes what is actually there. Recipe recommendations are built from the ingredients you already have, helping reduce food waste and the daily question of what to make.
  • AI Tailor
    Generate new recipes through search, and AI Tailor automatically caters them to your profile. You can also customize anything further, swapping ingredients, adjusting portions, or changing a technique to fit your kitchen.
  • AI Chef
    Cook hands-free with voice guidance that walks you through every step. Ask questions, request substitutions, or adjust on the fly through voice or text chat, with multilingual support across the languages offered.
From Idea to Launch

I built every part of Meal Melody. Here's what that looked like.

  • Product Manager
    Owned product strategy and Solution Intent for Meal Melody. Authored the PRD, defined the Vision and MVP scope, and maintained a WSJF-prioritized Program Backlog structured as Epics decomposed into Features. Sized work using Cost of Delay and Job Size inputs and set Program Increment Objectives with clear business value each iteration.
  • Product Owner
    Owned the Team Backlog and wrote user Stories to INVEST conventions with testable acceptance criteria. Enforced Definition of Ready at story acceptance and Definition of Done at closure, ran Backlog Refinement, Iteration Planning, and Iteration Retrospectives across every cycle, and ROAMed risks per iteration.
  • UX Designer
    Designed the complete UI in Lovable across the pantry and fridge scan flow, dashboard, recipe detail, hands-free cooking mode, onboarding, and dietary personalization panels. Built a design system with typography, spacing, and color tokens for cross-screen consistency. Applied mobile-first standards for the Capacitor launch, including 44 point touch targets, accessible contrast ratios, and native script support across six writing systems.
  • AI Product Lead
    Defined AI behavior and prompt strategy on top of Lovable's built-in Gemini access across image recognition, dietary recipe adaptation across 50+ cuisines, multilingual voice chat, and read-aloud cooking guidance. Wrote prompts as Stories with acceptance criteria on accuracy, tone, and latency. Ran variant testing on real recipes and beta users, and defined model fallback behavior for edge cases.
  • Full-Stack Engineer
    Built and debugged core Features using Claude Code, managed the GitHub repository with a structured branching workflow, and integrated Supabase for authentication, storage, and database. Handled the Lovable Publish gate on every frontend commit as part of the Continuous Delivery Pipeline.
  • QA Engineer
    Authored test plans covering the primary user flow, regression, edge cases, and Built-in Quality checks against the locked product rule. Validated each Lovable Publish by diffing shipped commits, sequencing tests by risk, and running regression smoke sweeps. Maintained coverage against a baseline of 1,376 tests across 91 files and added regression Stories when defects surfaced.
  • Growth
    Ran hypothesis-driven A/B experiments as Stories with pre-defined success metrics and minimum lift thresholds. Tested onboarding-to-dashboard flow variants, personalized AI prompt copy across beta segments, and AI Chef entry point placement. Tracked leading and lagging indicators, promoted winning variants into the base experience, and archived losers with post-experiment write-ups.
  • User Researcher
    Ran one-on-one interviews and structured surveys with 20+ beta users, feeding output into Story creation and Backlog Refinement. Synthesized qualitative and quantitative signal to inform Program Increment Objectives, driving a 45% lift in daily active usage and 30% reduction in load time through iterative releases.
Stack

Lean tools, end-to-end ownership.

Lovable Supabase Claude Code Gemini APIs React TypeScript GitHub

See it in action.

The site walks through the product. The app proves it works. Try Meal Melody yourself.

mealmelody.com