Light does not create.
It reveals.

Computer engineer from PUC Minas, with an MBA in Solutions Architecture from FIAP and a master's degree in Computer Science from UFSCar. For 21 years I have turned business problems into digital solutions, guided by a simple conviction: the best architecture is the one that solves the problem without creating new ones.

I have led multidisciplinary teams through the full lifecycle of digital solutions, from conception to delivery, at large corporations in Healthcare, Legal, Financial Services and Insurance. My work spans Cloud Computing, across migration and native, hybrid and multi-cloud environments, as well as Data Science, Artificial Intelligence, Distributed Systems and process automation.

My research is in artificial intelligence for medical imaging. My master's work addressed multi-pathology segmentation in lumbar spine MRI. My PhD investigates reliability in medical visual question answering, so that a model's answer genuinely depends on the evidence present in the image.

I teach in FIAP's MBA+ program and in PUC Minas' graduate program, covering Data Architecture, Data Streaming, Computer Vision, API Management and Multimodal Artificial Intelligence. In the classroom and on projects I make the same move: connecting the technical decision to the business outcome.

What drives me is the moment a complex system finally becomes clear.

My PhD investigates reliability in Medical Visual Question Answering, or Med-VQA, in the Graduate Program in Computer Science at UFSCar, Sorocaba campus, advised by Jurandy Gomes de Almeida Junior.

The problem is easy to state and hard to solve. A model can answer a clinical question about a scan correctly without having used the image at all, leaning on statistical shortcuts in the language. A right answer for the wrong reason does not support a clinical decision.

The proposal moves supervision from the discrete space of language to a continuous latent space, and treats four requirements as interdependent: answer adequacy, genuine use of visual evidence, anatomical grounding, and robustness under distribution shift.

My master's work, also at UFSCar, addressed multi-pathology segmentation in lumbar spine MRI.