Designing an app to help Algerian farmers choose the right crops using Machine Learning and mobile-first UX.
Master's project (Solo)
Product Design & Development
Android, iOS
Figma, Flutter, Python, TensorFlow
In rural Algerian communities, farmers often struggle to choose the right crops due to limited access to agricultural advisors and reliable data. I designed and developed a smart farming app that helps farmers get real-time soil classification and crop recommendations using Machine Learning and a mobile-first user experience.
This project was initiated as my final-year Master's project with the goal of creating a practical solution for farmers who lack access to agricultural expertise and modern farming technologies.
To understand the real challenges farmers face, I interviewed 7 farmers and 2 agricultural experts in rural Algeria. Each had different farming backgrounds and experiences, but all shared similar frustrations with accessing reliable agricultural information.
Two distinct personas emerged from research: the experienced Rural Farmer and the tech-savvy Young Farmer.
Mohamed has farmed for decades using traditional methods. He owns a smartphone but uses it mainly for calls. He struggles with complex apps and needs large buttons and simple instructions.
Karim has a small farm and is tech-savvy. He uses smartphone apps daily and wants precise, data-backed recommendations to optimize his farming operations.
To create a solution that addresses both personas' needs, I developed a mobile app with a tiered interface and integrated machine learning capabilities.
Algerian farmers need a way to get accurate crop recommendations based on their soil conditions without requiring technical expertise or constant internet access.
Create a mobile app that provides real-time soil classification and crop recommendations using ML, with an interface accessible to farmers of all technical backgrounds.
Optimized for smartphone use in field conditions with large touch targets and simple navigation.
Capture soil photos for instant classification using computer vision algorithms.
Core functionality works without internet connection in remote farming areas.
Recommendations tailored specifically for Algerian climate and soil conditions.
I started with low-fidelity sketches focusing on the core flow: capturing soil data and receiving crop recommendations. The design prioritized large touch targets, minimal text, and clear visual hierarchy.
After validating concepts with potential users, I created high-fidelity prototypes in Figma and conducted usability tests with 5 farmers, iterating based on their feedback.
The final app was developed using Flutter for cross-platform compatibility, TensorFlow Lite for on-device ML, and Firebase for backend services. The solution was tested with 12 farmers over a 3-week period.
Research & User Interviews
Design & Prototyping
ML Model Development
App Development & Testing