Lucas Garcia de Souza

Scientific Initiation Fellow in Artificial Intelligence

Applied conversational-AI research for searching, retrieving and analyzing intellectual-property assets, combining distributed applications and response evaluation.

• Integration across Java/Spring Boot, Python/FastAPI and Angular.
• RAG pipelines, BM25, embeddings, prompts and LLM provider adapters.
• Experiments with evaluation datasets and relevance, faithfulness and match metrics.

Software Quality and Testing at LEDS

Volunteer participation in a real software project, covering functional testing, requirements validation, defect investigation, test-data preparation and evidence documentation.

• Audit covering 88 considered tests and 87 recorded evidence items.
• 78 tests passed, 8 failed and 1 was inconclusive.
• Direct coverage of 14 of 28 functional requirements; 5 were blocked.

Income Predictor (Undergraduate Capstone)

Data product using PNAD Continua microdata, feature engineering, comparisons across econometric and machine-learning models, a FastAPI API and a React interface.

• Hybrid model with approximately 0.49 R² on the original income scale.
• Documented preparation, training, evaluation and inference pipeline.

International Academic Mobility

Selected for a study period at IPB without a mobility grant. The experience remains described as forthcoming until it formally begins.

BSc in Information Systems

Education in software engineering, databases, web development, artificial intelligence, machine learning, architecture, testing, research and extension. Reported academic score: 88.26.

Technical Diploma in Internet Computing

Technical foundation in programming, web development, logic, databases and internet technologies.