Postgraduate professor at the Data Mining and Big Data course of Instituto Mauá de Tecnologia.

Senior machine learning engineer focused on recommendation algorithms, engineering team management. Responsible for team management and value delivery with fast and reliable delivery.

Machine Learning Engineer responsible for developing the product recommendation tool for client convergence optimization. Interacting with several different clients working with software and solution architecture.

Machine learning engineer responsible for creating and maintaing the logistics algorithm capable of coordinating and optimizing product distribution.

Machine learning engineer responsible for several AI projects focused on data gathering and security testing with AI within big farm industries.

Most dynamics systems can be easily simulated as a state-space linear system, by just multiplying a particular set of four matrices. The Giant Magellan Telescope is simulated by using a particular set of three matrices (A, B, C) from a finite element state model. But the problem in critical systems is that it is not always that simple... The problem of computing these matrices is that the Giant Magellan Telescope is a very complex system and therefore has very complex/big matrices. Numerically, it can get to something close to 11000 states, 20000 outputs and 8500 inputs, i.e. matrices with 11000x11000, 20000x11000 and 11000x8500, dimensions to be computed in a rate of 2000 Hz. Common simulation software cannot handle this type of complexity. With that in mind, one needs a designed simulator to compute the telescope dynamic model. The task provided was to develop this simulator with performance close to real-time. For that, a GPU CUDA based parallel computing simulator was developed mostly with python and C++ to get the desired performance.