At DataAnnotation, I engineer multi-turn conversational datasets for RLHF and SFT on U.S.-based foundation model programs. I also evaluate complex Python coding and reasoning tasks to identify failure modes, edge cases, and reliability risks.
At EPM - Empresa Paranaense de Montagens, I developed Python applications, Dockerized services, and AI-driven agents to automate enterprise workflows and document processing. I established technical architecture for automation and agentic AI workflows across the engineering organization.
At EPM, I built predictive models and data pipelines in Python and SQL to forecast project costs and optimize resource allocation, enabling $10 million in sales in 2025. I also conducted bottleneck analysis and collaborated on bringing data-driven applications into production.
I built Dark-Souls-App, a Python question-answering system using RAG and GraphRAG for retrieval over multi-relational entity graphs. My other projects include machine-learning pipelines, industrial electrical measurement automation, and an image-based nanoparticle quantification device.
