Nelson Joppi

Transformed geological data into insights for Petrobras, maintaining Petrel plugins and training ML models for advanced geological analysis. Delivered presentations showcasing ML applications to geologists.

Core developer of VigIA, an AI-driven platform that processes public expenditure data to detect procurement irregularities. Revamped AI architecture for various LLM applications and integrated ML models on GCP and Azure, enhancing audit efficiency and reducing costs.

Trained and benchmarked ML models on well-log data to predict lithology with ~80% accuracy, collaborating with geologists for feature engineering and model optimization.

Built responsive web dashboards and mobile apps using React and React Native, integrating with backend APIs under Scrum methodology.

Developed and optimized a genetic algorithm for urban road network design, performing detailed analysis and integrating heuristics to deliver a scalable solution.

My undergraduate thesis is on developing a poker-playing program that utilizes reinforcement learning techniques, search algorithms, and advanced game theory concepts.