This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend
What I Built
I built Wellness AI Lens specifically for my close friend, Aarav.
Aarav recently decided to take his fitness and dietary health seriously. However, he struggled constantly with two major friction points:
- Hidden Ingredients & Allergen Uncertainty: He suffers from recurring digestive sensitivity and lactose intolerance, making dining out or trying new meal options a game of nutritional roulette.
- Tedious Manual Logging: Mainstream fitness apps require typing out every individual item, measuring ounces, and guessing ingredients—a chore he gave up on after just three days.
Wellness AI Lens solves this by acting as a pocket health companion. Aarav simply takes or uploads a photo of his meal, and the application instantly:
- Identifies every visible component on the plate.
- Delivers an accurate macronutrient (protein, carbs, fats) and calorie breakdown.
- Scans for high-risk allergens and inflammatory ingredients tailored to his sensitivities.
- Recommends contextual post-meal activity and lifestyle adjustments to avoid glucose spikes.
What Aarav said when I showed him:
"I snapped a photo of a curry bowl at lunch, and within seconds it flagged hidden cream content I wouldn't have noticed. Not having to manually search an ingredient database to track my lunch makes me actually stick to my goals."
Demo
- Live Application: https://wellness-ai-lens.vercel.app
* Demo Video:
Wellness AI Lens: Scan Your Food with AI | Jayesh Jain posted on the topic | LinkedIn
Wellness AI Lens – 60-second demo
Scan Your Food: https://lnkd.in/dUfighAa
🚀 See how AI turns every plate into insight
🔍 Computer-vision scan → instant macros, micros & allergens
💯 Smart health score for each bite
🤖 Built-in chat dietitian for follow-up questions
📈 Tracks goals, meals, steps, sleep & more
🔔 Auto reminders, daily summaries, weekly reports
🔐 Secure auth + dark/light themes
Watch the quick video and imagine smarter eating made effortless.
linkedin.com
Wellness AI Lens: Scan Your Food with AI | Jayesh Jain posted on the topic | LinkedIn
Wellness AI Lens – 60-second demo Scan Your Food: https://lnkd.in/dUfighAa 🚀 See how AI turns every plate into insight 🔍 Computer-vision scan → instant macros, micros & allergens 💯 Smart health score for each bite 🤖 Built-in chat dietitian for follow-up questions 📈 Tracks goals, meals, steps, sleep & more 🔔 Auto reminders, daily summaries, weekly reports 🔐 Secure auth + dark/light themes Watch the quick video and imagine smarter eating made effortless.
linkedin.com
Code
The complete source code is public and open-source under the MIT License:
Jayesh-JainX
/
Food-Analysis-Platform
AI-Powered Food Analysis, Nutrition Tracking & Health Intelligence Platform
Wellness AI Lens - Food Analysis Platform
🌟 Overview
Wellness AI Lens is an advanced AI-powered food analysis and health management platform designed to help users make informed nutritional and lifestyle decisions. By leveraging modern computer vision and multimodal LLM intelligence via Hugging Face Inference API, the platform enables users to scan food items, retrieve accurate nutritional breakdowns, track fitness goals, explore disease management strategies, and access tailored exercise routines.
✨ Key Features
🍎 AI Food Analysis & Vision Scanning
- Image & URL Analysis: Analyze food items using camera uploads or direct image URLs.
- Nutritional Breakdown: Real-time identification of macronutrients (protein, carbs, fat) and micronutrients (calories, sodium, fiber, sugar).
- Health Score Algorithm: Automated health scoring based on nutritional density and ingredient composition.
- Allergen & Ingredient Detection: Highlighting potential allergens and health advisories.
- Scan History &…
How I Built It
Wellness AI Lens is architected to be responsive, scalable, and built around open-weight intelligence:
- Frontend: Built with Vite + React for an ultra-fast, responsive mobile-first UI.
- Backend: A Node.js server handling image pre-processing, prompt structure, and authenticated endpoints.
- Authentication & Database: Supabase handles secure user authentication and persists historical food scans, allergen profiles, and fitness logs.
- Hosting: Deployed on Vercel for low-latency serverless edge delivery.
-
Open-Source AI Core: The visual inspection pipeline is powered by the open-weight multimodal model
meta-llama/Llama-4-Scout-17B-16E-Instruct(accessible via the Hugging Face Inference API / Novita provider).- The image payload and a domain-specific structured prompt are dispatched to the model.
- The model outputs a strict JSON payload containing dietary tags, detected allergens, estimated macro distributions, and personalized advice based on user profile flags stored in Supabase.
Why Does Open Innovation Matter?
Health, diet, and lifestyle data are among the most sensitive personal records an individual generates daily. Building Wellness AI Lens on an open foundation was an intentional choice for three reasons:
- User Privacy & Sovereign Health Data: Closed commercial vision APIs (like proprietary frontier models) retain user data according to opaque corporate terms. By building on open weights via Hugging Face endpoints, we retain full ownership of our pipeline. Because the model weights are public, this architecture can be ported to run on local hardware (e.g., self-hosted Ollama on a home server) without rewriting a single business-logic prompt.
- Protection Against Vendor Lock-in & Token Extortion: Proprietary vision APIs are costly for daily nutritional tracking where a user scans 3–5 meals a day. Open-weight models democratize access—allowing independent developers to offer high-caliber multimodal analysis without incurring prohibitive per-token taxes.
- Reproducibility & Customization: Open weights allow the developer community to audit model bias, fine-tune for regional cuisines (such as Indian and Asian regional dishes, which proprietary Western models often misclassify), and run lightweight quantized checkpoints on commodity hardware.
Open innovation ensures that tools designed to help people live healthier lives remain accessible, auditable, and free from corporate walled gardens.
Top comments (1)
Multimodal is the right call for food, typing out what is on a plate is way more friction than just pointing a camera at it. Nutrition is also one of those areas where generic advice is useless, good that it is built for one specific friend.