Hello, I’m João.

I’m a PhD candidate in Electrical Engineering at EPFL, based at the Idiap Research Institute in Martigny, Switzerland, working in collaboration with Honda Research Institute Europe. My research is on neuro-symbolic AI for reliable reasoning: making fluent neural and language models dependable on hard, structured problems by coupling them to symbolic structure that is sound, checkable, and interpretable.

My main application is engineering design, specifically the automated synthesis of mechanisms. I chose it because it is provably hard and comes with a deterministic verifier, so “did the model actually reason correctly?” becomes a measurable question. The areas I work in most are neural-symbolic integration, symbolic reasoning and formal verification, and agentic reasoning.

Before EPFL, I was a research intern at IBM Research (Rio de Janeiro), building LLM pipelines for scientific knowledge extraction, and earned my MSc and BSc at the Federal University of Rio de Janeiro (UFRJ). I’ve also taught programming to 100+ undergraduates and spent several years as a software engineer building applications used at scale.

My long-term goal is to develop AI methods and systems that accelerate engineering innovation while preserving the mathematical rigour, reliability, and safety that real-world design demands.