STEM Data Annotation Expert / Head of Engineering Budget Section (Brazilian Air Force)
Led the structured classification, evaluation, and validation of 1,400+ technical engineering submissions, utilizing frameworks analogous to data labeling workflows. Designed and implemented taxonomy, label schema, and quality-control processes for complex STEM datasets to ensure accuracy and reproducibility. Performed step-by-step reasoning trace audits, identifying logical errors and documenting explicit assumption chains for fine-tuning datasets. • Developed gold-standard technical assessments and reference solutions suitable for AI training data • Decomposed multi-constraint problems to produce verifiable Chain-of-Thought annotation • Reviewed and rated model outputs for logical integrity, ambiguity, and adherence to instructions (RLHF) • Applied quality control, inter-annotator agreement, and feedback loops to maintain high annotation standards.