I'm a Software Engineer bridging scalable backend systems and applied AI to deliver measurable impact. Currently building production systems at Livvie (Miami) and conducting NLP research at TAIL/UFPB.
My work spans three areas:
- Applied AI — LLMs, RAG pipelines, NLP for social impact
- Data Engineering — large-scale pipelines, AWS data lakes, Bronze/Silver/Gold architectures
- Backend — Clean Architecture, performance optimization, distributed systems
Currently pursuing a B.Sc. in Data Science for Business at UFPB (2024–2027).
I'm drawn to the intersection of rigorous mathematics and real-world impact — the kind of problems where understanding the theory actually changes what you build.
Natural Language Processing — political bias in LLMs, rhetorical manipulation detection, low-resource language modeling. Language is messy and political; I like that.
Computer Vision — representation learning, geometric deep learning, how spatial structure emerges from optimization. The connection between convolution and translation equivariance still feels elegant every time.
Robotics — perception pipelines, sensor fusion, the feedback loop between physical systems and learned models. Where math meets the physical world most directly.
What ties these together is the math underneath. I find it genuinely fascinating how differential equations, symmetry groups, and information geometry keep showing up across seemingly unrelated fields — and how that understanding shapes better engineering decisions.
Physics is a constant reference point for me: not just as inspiration, but as a way of thinking about systems, constraints, and what it means for a model to actually understand something.
AI & Data Science
Backend & Infrastructure
Data & Databases
I'm an NLP Researcher at TAIL (Technology and Artificial Intelligence League — UFPB), focused on:
- Political bias detection in large language models
- Rhetorical manipulation in Brazilian parliamentary discourse
- Corpus annotation and benchmark construction for PT-BR NLP
"Reliable AI starts with solid foundations — rigorous EDA, honest evaluation, and reproducible methodology."