Gabriel Rovina

Coded the infrastructure behind VTEX IO, a cloud-based serverless platform for web development. Spent my time designing the applications and improving the resiliency of the services (Kubernetes distributed and large-scale). Worked with Golang and C#, as well as Redis and deployment in multiple cloud platforms.

Researched, designed and deployed AI-based products for vision: fully integrated systems for facial recognition, object detection, crowd counting and so on.

Implemented a TensorFlow-based deep learning library from scratch, then used it to combine multiple neural network architectures. The models processed stock market data. Also created numerous performance metrics and visualisation tools to evaluate and diagnose results from these models. "Walnut Algorithms is a technology firm focused on applying the latest advances in data science and machine learning research to the financial markets."

Built patient-specific 3D models of the heart from MRI data, ran finite element simulations of the electrical activity of the heart and optimised pacemaker setup for cardiac resynchronisation therapy. Better positioning and timing allow more efficient therapies, considerable improvement in patient's quality of life and reduction hospitalisation rate. Reduced simulation runtime to less than 37% of original time. Supervised by Luca Formaggia and Christian Vergara.

Developed a recommendation system for adaptive learning, based on own graph modelling of all the user-system interactions, similarity metrics between users and ant colony optimisation to find the most fruitful paths for each user, all implemented from scratch. Performed several analyses on usage data and created a web platform for report generation, interaction with results and usage summaries dashboard.