I’m a Brazilian electrical engineer working towards my Ph.D. under the supervision of Luis A. Aguirre. My current research interests lay in the intersection of machine learning, signal processing and control theory. I’m particularly interested in training models of dynamic systems from observed data and on the development of new methods to cope with this kind of problem.

I’ve work on research problems such as: the comparison between different training methods for neural networks dynamic models, the estimation of output error models using multiple shooting and the development of methods for choosing data windows to train empirical models. The latter was my contribution to a R&D project with Petrobras Oil Company.

I’ve contributed to large open-source projects. More specifically, I am a member of Scipy core development team, having contributed to optimization and signal processing packages with the implementation of IIR filters (implementation of functions iirnotch and irrpeak), trust-region optimization methods (implementation of methods trust-exact and trust-constr) and other general improvements (such as the implementation of quasi-Newton strategies BFGS and SR1).

On the industry side I’ve worked at Invent Vision, a startup that provide computer vision solutions to industrial applications. I was part of the hardware development team and worked designing FPGA-based cameras. The major project I have worked on while there was the design and implementation of a stereo camera.

I have also successfully completed Google Summer of Code program. My project was the development of an interior-point solver for large-scale nonlinear programming problems. The implementation is available in this github repository and have been integrated to SciPy library (release 1.1 or above) under the name of trust-constr.

I’m comfortable working with C/C++, Python, Matlab, Julia and R and I have skills on optimization, machine learning, system identification and signal processing.