Egonym AI

In today’s digital age, privacy has become a luxury when it should be a basic right. Egonym addresses this challenge by providing businesses with sophisticated visual anonymization technology that ensures data protection compliance while maintaining data integrity for downstream applications like sentiment analysis, demographic profiling, and ML training.

Country:
Switzerland
Industry:
AI
Technologies:
Django, Python, React, FastAPI, PostgreSQL, Redis, Celery, Docker, Machine Learning

Egonym is a Swiss startup that develops AI-based privacy enhancing technologies. Their flagship product is a cloud-based software solution that anonymizes facial identities in images and videos while preserving key data attributes like age, demographics, and sentiment for business analytics and ML training purposes.

Tasks

We developed the core distributed system architecture and implementation of the web application:

  • Design and implementation of scalable distributed architecture separating web operations from ML processing
  • Development of secure media upload and processing pipeline
  • Implementation of asynchronous task management system for handling ML operations
  • Creation of isolated data storage patterns for ML operations with centralized data management
  • Development of authentication and session management system
  • Implementation of API endpoints for ML model serving
  • Infrastructure setup including containerization and security measures
  • Integration of ML processing service with main application server
  • Implementation of temporary storage solutions for ML processing data
    • ‍

Team set-up
Our team focused on building the core infrastructure and web application:

  • Backend developer (Django, FastAPI)
    • System architecture + DevOps (AWS, terraform, terragrunt, Github CI/CD)
  • Frontend developer (React)

Results & Benefits

The development resulted in a fully functional web application with these key achievements:

  • Successful implementation of distributed architecture enabling ML processing
  • Secure and efficient media processing pipeline for image and video anonymization
  • API-ready system allowing for future integration with client systems
  • Robust task queue system handling asynchronous ML operations
  • Clear separation between web and ML services ensuring system reliability
  • Support for multiple file formats (JPEG, PNG for images; MP4 for videos)
  • Foundation for future expansion into real-time processing capabilities

‍
The system has become ready for client onboarding and handling the processing requirements of various use cases including marketing, retail analytics, and ML training data preparation, while ensuring privacy compliance.

Codepole helped with the launch of our current platform. Implemented features and products enabled us to rapidly scale our existing Swedish business and expand to new markets.

Robin Tjassens, Kompar

Codepole managed to provide me with top quality candidates in short period of time. Their support and hands-on approach throughout the recruitment process were invaluable.
‍

Daniel Abebe, HuggyStudio