MenuRX: AI Diets & Clinical Solution

Overview MenurX is a comprehensive mobile and web platform designed to transform how individuals with specific dietary and clinical needs read restaurant menus. Acting as an intelligent nutrition assistant, it translates listed ingredients into clear, goal-aware guidance without hallucinating "off-menu" items, empowering users to make safe, informed dining choices.

The Challenge Navigating restaurant menus with clinical dietary constraints (e.g., low-glycemic, anti-inflammatory, low-histamine) is full of unknowns. Vague menu descriptions for sauces or marinades often leave diners guessing, and standard AI solutions tend to invent ingredients that aren't actually written, creating potential health risks.

The Solution I engineered a robust, AI-powered ecosystem that analyzes scanned or pasted menus to deliver deterministic dietary compatibility scores. If a dish's components are underspecified, the app intentionally surfaces a "needs verification" state, equipping the user with exact questions to ask the restaurant staff. The platform provides actionable modifications and safe alternatives based strictly on what is written on the page.

Technical Architecture & Engineering The system architecture was designed for high performance, absolute data accuracy, and seamless cross-platform scalability:

Cross-Platform & Native Frontend: Built a scalable, responsive interface with a sleek, modern UI/UX using a clean, custom Flutter codebase for complete web and mobile parity. To handle specialized hardware interactions and highly specific edge cases, I developed and integrated custom flows utilizing Native Swift (iOS) and Kotlin (Android).

AI Intelligence & Orchestration: Powered the core menu-parsing engine using Google Gemini. I utilized Firebase Genkit and Firebase AI logic to orchestrate OCR data extraction, accurately mapping dish names and ingredient phrases exactly as written.

Clinical Rule Engine: Developed and integrated custom ML models specialized in medication and health data. This ensures that extracted ingredients are evaluated deterministically against a curated clinical knowledge base for various diet lenses (Keto, Gluten-Free, etc.) without generative guesswork.

Backend & Cloud Infrastructure: Architected a robust, real-time backend using Firebase to handle secure user authentication, rapid data synchronization, and scalable cloud performance across both web and mobile environments.

Intelligent Menu Capture: Scans physical menus via camera or parses text from PDFs and websites, cleaning up OCR quirks on the fly.

Strict Ingredient Extraction: Breaks down dishes into explicitly named components—no silent add-ons or assumed ingredients.

Goal-Aware Scoring: Delivers a cautious score (or a verification prompt) for selected health goals with plain-language, logic-based explanations.

Actionable Modifications: Generates practical asks for the kitchen and highlights safer alternatives directly from the same menu.

Outcome A highly reliable, educational tool shaped by clinical rigor and advanced AI logic, delivering a secure, transparent dining assistant that translates complex menu data into everyday practicality.