Loading project case study.
Back to selected work
AgriTech + ML
Goat Master
A herd management app that combines animal records with photo-based identification and support for offline field work.
- Project type
- Herd management app
- Industry
- AgriTech + ML
- Core stack
- Next.js / React / TensorFlow.js / FastAPI
Problem
The challenge.
Keeping track of individual animals is difficult when goats look alike and health records are spread across notebooks or spreadsheets.
Build
What I built.
I built herd records for health, ownership, and productivity alongside a photo matching workflow. The project combines browser-based image analysis, a Python ML service, and vector search.
Outcome
What it supports.
The app is designed to help a farmer find an animal from a photo and keep its records available in the field.
Highlights
Inside the project.
- Browser-based image analysis for field use.
- Server-side image analysis and similarity search.
- Animal records for health, ownership, and productivity.
- Local storage for farm work with an unreliable connection.
Technical stack
Tools and technologies.
- Next.js
- React
- TensorFlow.js
- FastAPI
- YOLOv8
- ResNet50
- PostgreSQL
- pgvector
Design challenges
Details that shaped the work.
Animals can look similar, and farm photos vary in lighting and angle. The matching workflow has to account for those differences.
The app separates browser-side analysis from server-side matching and stores farm records locally for use with an unreliable connection.