Viderai

In early 2020, the greenfield R&D startup project Viderai was born, with the ambitious goal of developing Computer Aided Decision software for detecting diabetic retinopathy from retinal scans. I have been part of this project for 7 years.

As an ML Engineer, I built numerical models and predictors from scratch using the GPU superclusters Barbora and Carolina. Later, I refactored the proof of concept into a distributed modular monolith capable of horizontal scaling, opting for Microsoft Orleans and actor model over microservices.

I have become a strong advocate of distributed modular monoliths for software used as medical devices. I love their ease of maintenance, cost-effectiveness, simplicity, security, and robustness, which makes them capable of handling thousands of concurrent users. However, I understand that they are not the perfect fit for every use case, and the right architecture always depends on the specific needs of the project.