• Computer Science BSc @ University of Brasília (UnB), 2025
• 6 years of experience in applied AI projects, 4+ in consulting across finance, education, energy, and infrastructure
• Specialized in LLMs, RAG systems, and AI agents, with projects in production
• Author of DiggVis - visual analytics system for public contract auditing, awarded maximum grade (SS) at UnB

AI & LLMs: Python • LangChain • LangGraph • LlamaIndex • OpenAI GPT • RAG • AI Agents • Prompt Engineering • Guardrails • Fine-tuning • Hugging Face
NLP & Extraction: NER • Text Classification • Summarization • Semantic Embeddings • Vector Search • OCR
MLOps: MLflow • CI/CD • Automated Testing • REST APIs (FastAPI • Flask)
Data & Cloud: SQL (PostgreSQL • MySQL • SQL Server) • AWS (Lambda, S3, EC2, RDS) • Azure • Terraform • Docker • Git
Vector DBs: FAISS • ChromaDB

• Deploying LLMs and RAG systems in real production environments
• Building an intelligent tutoring system with RAG, adaptive question recommendation, personalized feedback, and guardrails
• Designing AI agents for document extraction, classification, and automated decision-making
• Developing large-scale document processing pipelines with OCR, NER, and confidence scoring

• DiggVis: Visual analytics for public contract auditing using LLMs, semantic embeddings, and interactive clustering. Undergraduate thesis awarded maximum grade (SS) at UnB.
• Joana: RAG-based job analysis tool that evaluates job descriptions against a resume, identifying match strength, skill gaps, and overall role alignment.
• TatikToken: Custom Byte Pair Encoding (BPE) tokenizer built from scratch, inspired by GPT architecture, with parallel processing support.