Inspiration
Food insecurity is a daily challenge in regions like Haiti, where people often have to make meals from whatever is available. We wanted to empower users to create nutritious, fuel-saving recipes from any ingredients, even with just a photo.
What it does
ManjeAI lets users type or photograph available foods, detects ingredients, and generates a complete recipe with nutrition breakdown and local, context-aware suggestions. All recipes are saved for easy reuse.
How we built it
We used Flask (Python) for the backend, Google Gemini for recipe and nutrition generation, LogMeal for ingredient detection from photos, SQLite for history, and a modern mobile-first UI with Tailwind CSS, Chart.js, and marked.js.
Challenges we ran into
Ensuring accurate ingredient detection from diverse food photos, parsing AI output reliably, and making nutrition and suggestions context-aware for different cuisines and dish types.
Accomplishments that we're proud of
A seamless, mobile-first experience that works with both text and photos, context-aware nutrition suggestions, and a beautiful, easy-to-use interface.
What we learned
How to combine multiple AI APIs, robustly parse and visualize AI output, and design for real-world constraints in food-insecure regions.
What's next for ManjeAI
Add offline and SMS/WhatsApp support, expand local language options, partner with local organizations for ingredient data, and test with real users to maximize impact.
Log in or sign up for Devpost to join the conversation.