Luiz Ferreira

Build and operate an award-winning Python/Django RAG troubleshooting solution for one of Latin America's largest companies, achieving ~85% successful resolution within three interactions, ~10% safe abstention, and ~5% hallucination. Work in a client-facing consulting environment, translating workflows and production evidence into system requirements and architecture changes.

Evaluate LLM outputs across reasoning, factual accuracy, instruction adherence, and safety, producing calibrated assessments used in model benchmarking and reliability improvement. Analyze ambiguous and adversarial cases to separate factual support, reasoning validity, grounding, and instruction compliance.

Supported public procurement, administrative, and financial workflows by reconciling government portals, accounting records, regulations, and internal sources into actionable documentation and reporting. Improved traceability and decision support by organizing calculations, regulatory references, source evidence, and supporting documentation for time-sensitive public processes.

Pursuing a Technology Degree in Systems Analysis and Development, expected to complete in November 2026.

Completed a Technical Diploma in Business Administration from 2011 to 2014.