Jan 2026 – Mar 2026
Culinara
AI-Powered Recipe & Pantry Platform
Overview
A full-stack AI recipe platform where users photograph their pantry and receive step-by-step recipes instantly — with pantry management, PDF downloads, and 92% ingredient detection accuracy.
Problem Statement
People waste food because they don't know what to cook with what they have. Traditional recipe apps require you to already know the ingredient names.
Key Highlights
Image-based ingredient detection via Google Gemini AI with 92% accuracy in user tests across camera, upload, and drag-and-drop inputs
Full-stack platform with Next.js + Strapi backend, cutting recipe discovery time by 60% vs browsing
Secure auth and API fallback handling on Vercel with page load times under 1.5s across all tiers
Architecture
Next.js App Router frontend with Strapi CMS backend. Google Gemini Vision API processes ingredient images; results feed a recipe generation prompt pipeline.
Challenges
Handling fallback gracefully when the Gemini API is rate-limited, and building an ingredient detection UX that works reliably on low-quality phone photos.
Learnings
Multimodal AI integration, headless CMS architecture with Strapi, and production-grade error handling for third-party API dependencies.