AI Analytics Bottom Page
Daily Tracker Page
Landing Page
AI Analytics Page
Dashboard Page
Meal Planner Page
Recipes Page

proFoodie – Building AI-Powered Nutrition From My Own Weight Loss Journey

What Inspired Me

In early 2023, I weighed 88 kg and was struggling with my nutrition. Like many people, I was overwhelmed by vague advice, inconsistent tracking apps, and conflicting diet tips online.

Then AI exploded — tools like ChatGPT, Midjourney, and other LLMs were changing everything.
Naturally, I asked myself:

“What if AI could understand food just like it understands language?”

That's when I began experimenting with AI for nutrition, using models to analyze meals and build personalized meal plans. By mid-2024, I had lost 17 kilograms through AI-assisted food planning.

People around me started asking, “How did you do it?”

That question sparked proFoodie — a smart, global-first nutrition platform powered by AI.

How I Built It

proFoodie is a full-stack AI-powered platform built with:

🧠 AI Models: Computer vision + LLMs for food identification and nutrition breakdown
🌐 Frontend: React + Tailwind in a universal PWA shell (offline-first)
📦 Backend: Node.js + Python APIs for inference and data management
🧠 Nutrition Intelligence Engine: A fusion of food science + cultural understanding
🌍 Globalization: Supports 50+ cuisines and cultural portion sizes using a dynamic metadata engine

I also implemented a global AI function:

const analyzeGlobalCuisine = async (image, description, userCulture) => {
  const culturalContext = getGlobalCuisineDatabase(userCulture);
  const aiEstimate = await universalNutritionAI.analyze({
    image,
    description,
    cuisineType: detectCuisineType(image, description),
    culturalPortions: culturalContext.servingSizes,
    regionalIngredients: culturalContext.commonIngredients
  });

  return culturallyAwareNutritionData;
};

What I Learned

Cultural food data is hard to find – most nutrition platforms ignore cultural nuance.
Accuracy matters more than features – people want trust in nutrition data.
AI needs human reinforcement – I built feedback loops into the app so users can improve recognition.
User retention is everything – I realized the app has to feel intuitive for a daily habit to form.

Challenges I Faced

Food Recognition Models: Getting models to recognize diverse cuisines was tough.
Cultural Bias: Most datasets are Western-centric. I had to build a small dataset of my own to support African, Indian, and Asian cuisines.
Offline Mode: Making the app work offline-first while keeping AI components functional was one of the biggest challenges.
Scaling AI on Budget: GPUs are expensive. I used quantization and model distillation with TensorFlow Lite to run inference affordably.

Math Behind My Journey

Target weight loss:

$$ \Delta W = W_{start} - W_{target} = 88\,\text{kg} - 71\,\text{kg} = 17\,\text{kg} $$

Caloric deficit needed:

$$ \text{Total Deficit} = 17\,\text{kg} \times 7700\,\text{kcal/kg} = 130,900\,\text{kcal} $$

Daily deficit for 6 months:

$$ \frac{130,900}{180} \approx 727\,\text{kcal/day} $$

AI helped me maintain a consistent 700–800 kcal/day deficit by optimizing meal plans without starving or guesswork.

Closing Thoughts

proFoodie is more than a product. It’s a mission born from my own health journey — built with code, vision, and empathy.

AI helped me lose weight. Now, I’m using AI to help the world eat better — one dish at a time.

“If AI can understand food like humans, maybe it can finally help us eat like we should.”

Whats next for proFoodie

At proFoodie we are building a universal health interfcase powered by AI. It’s a mission born from my own health journey.

Nutrition is deeply personal. Culture is non-negotiable. AI is the bridge.

“That’s the future we’re building with proFoodie.”

Tech Stack & Architecture

proFoodie is built as a modern, modular, and offline-capable frontend application using a robust and scalable web stack:

Frontend Framework

React 18.3 – Fast, dynamic UI with component-driven design and Hooks
Vite 5.4 – Lightning-fast dev server and build toolchain for React projects
TypeScript 5.5 – Static typing for safer code and faster developer iteration

Styling & UI

Tailwind CSS 3.4 – Utility-first CSS framework for responsive, clean design
Framer Motion 10.16 – Smooth animations and transitions for great UX
Lucide React 0.344 – Scalable icon system for consistent visual design

Data Visualization

Chart.js 4.4 + react-chartjs-2 5.2 – Beautiful and interactive data visualizations
Date-fns 2.30 – Lightweight date utilities for tracking nutrition over time

UX Feedback & Delight

React Hot Toast 2.4 – Toast notifications for feedback and alerts
React Confetti 6.1 – Lightweight celebration effects for user goal milestones

Tooling & Dev Experience

ESLint 9.9 + TypeScript ESLint 8.3 – Code linting and static analysis
PostCSS + Autoprefixer – CSS transformation and compatibility handling
React Refresh + ESLint Plugin Hooks – Fast refresh and React-specific linting