RESEARCH • ACADEMIC • PEER-REVIEWED
Research
Peer-reviewed publications, preprints, patents, and talks from the lab.
PREPRINTS
Preprints
This paper explores the emerging symbiosis between human developers and AI assistants, analyzing how this collaboration is redefining the software development process and the implications for human cognition.
An analysis of commit patterns in Git repositories to understand developer cognitive behaviors, identifying patterns that correlate with productivity and code quality.
Bachelor's thesis (Computer Science, IFSULDEMINAS). Presents the design, development, and evaluation of Cidadão.AI — a platform with 17 specialized agents using NLP, multi-agent architecture, and LLMs to provide conversational access to Brazilian government transparency data. Written in Portuguese.
AWARDS · RECOGNITION
Awards & Recognition
Academic and industry distinctions. This is the first page of a journey just getting started.
Undergraduate Thesis Competition in Information Systems
Cidadão.AI won 2nd place at the VIII CTDG-SI 2026, the Brazilian Computing Society competition recognizing the best undergraduate thesis projects in Information Systems. Oral defense during the 22nd Brazilian Symposium on Information Systems in Vitória/ES, with a 31.2% acceptance rate (10/32 entries).
One of 5 national finalists
Cidadão.AI was one of the 5 national finalists of the SBC Innovation Seal 2026, a Brazilian Computing Society competition recognizing Computing projects with the greatest potential for innovation and social impact. The final stage was an in-person pitch with jury evaluation at Computec, during CSBC 2026 (the largest Computing event in Latin America), in Gramado/RS.
First public distinction. More to come.
PATENTS
Registered Patents
Cidadão.AI — Multi-Agent Artificial Intelligence System for Government Transparency
Anderson Henrique da Silva · September 4, 2025
Multi-agent AI system for Brazilian government transparency — 17 specialized agents over 30+ federal APIs, with semantic routing, contract classification, anomaly detection via FFT spectral analysis, and three-tier cache RAG.
Registration: BR512025004322-8