Roman

Roman is a Solutions Architect at NVIDIA. He focuses on accelerating data analytics workflows and building better text datasets for large language models.

When scarce data blocks non-English AI development, Roman turns to GPU acceleration to speed up corpus curation. His work covers everything from basic text cleaning to semantic deduplication and synthetic data generation. He aims to make language technologies more inclusive and functional across languages like Spanish and French.