Andrea Gemelli

I am an AI Researcher & Engineer based in Paris, currently working on multi-agent architectures and on retrieval and orchestration pipelines underneath them.

I held PhD in Artificial Intelligence and three years in industry building around post-training and evaluation pipelines for LLM and agentic-based systems. My career, or CV for the details.

Open contributions: Doc2Graph framework for several document understanding tasks, Baguettotron-VLM, a sub-1B vision-language model and DocVQA datasets and models available on Hugging Face.