Nicolo Brandizzi
Research Scientist @Fraunhofer IAIS
Signup · Get unlimited contacts
WORK HISTORY
Research Scientist @Fraunhofer IAIS
Bonn, DE
Research and development in data pipelines for large language models (LLMs), with a focus on federated data privacy and data quality for multilingual models: Developing scalable data pipelines to optimize LLM training on HPC systems. Standardizing workflows and establishing best practices for AI data governance. Implementing dataset quality methods and advanced filtering to improve training reliability. Coordinated with project partners, representing technical work through publications, reports, presentations, and documentation. Prototyping agentic/multi-agent systems for collaborative and adaptive AI behavior. Leading on AI data governance: GDPR compliance, reproducibility standards, and metadata chains: EuroLingua: Data processing, multilingual dataset ablations. TrustLLM: GDPR compliance and metadata roadmap for responsible LLM use. OpenGPT-X: Data quality estimation, large-scale filtering, and pretraining pipelines. DKZ2R: Lead for reproducibility in data pipelines; organized hackathons and trainings. TAILOR Network: Organized workshops on trustworthy AI and multi-agent systems. Jupiter AI Factory: Fraunhofer lead, building infrastructure for European foundation models.
EDUCATION
Sapienza Università di Roma
Doctor of Philosophy - PhD, Computer Science
University of Groningen
Master's degree, Artificial Intelligence & Robotics
Sapienza Università di Roma
Master of Engineering - MEng, Artificial Intelligence And Robotics
Sapienza Università di Roma
Bachelor's degree, Mechatronics, Robotics, and Automation Engineering
ABOUT NICOLO BRANDIZZI
I am an AI Research Scientist at Fraunhofer IAIS, working at the intersection of large-scale language model training, reinforcement learning, and AI governance. My expertise covers the full pipeline: Dataset preparation & quality assurance for LLM pretraining and finetuning, ensuring scalable and trustworthy data workflows on HPC systems. Reinforcement learning, including RLHF and multi-objective finetuning, developed through my Ph.D. and applied across industry and academic projects. Agentic systems, where I have built multiple prototypes integrating multi-agent reinforcement learning and communication models. I also contribute to European AI initiatives such as OpenGPT-X, EuroLingua, and JUPITER AI Factory, coordinating data governance and reproducibility efforts.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.