Summary
Andrés Marafioti is a Multimodal Research Lead at Hugging Face with 11 years of experience building efficient, open, and deployable AI systems across vision, speech, and robotics. He holds a PhD in applied machine learning focused on generative audio and has published at top venues while translating research into lightweight models like SmolVLM and NanoVLM for on-device inference. Andrés co-developed SmolDocling with IBM and is extending multimodal research into robot instruction-following with SmolVLA, emphasizing memory-efficient, reproducible solutions. His background spans industry and academia—from surgical-robot perception at Intuitive to production ML at Unity—bringing a rare mix of signal-processing rigor and practical engineering. A prolific open-source contributor, he prioritizes community-driven toolkits that make state-of-the-art multimodal AI accessible beyond large labs.
11 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy - PhD, Electrical and Electronics Engineering, Doctor of Philosophy - PhD, Electrical and Electronics Engineering at Universität für Musik und darstellende Kunst Graz
Sound Engineer, Acoustics, Sound Engineer, Acoustics at Universidad Nacional de Tres de Febrero
Ingénieur en acoustique et vibrations, Mechanical Engineering, Ingénieur en acoustique et vibrations, Mechanical Engineering at ENSIM Ecole d'ingénieurs Le Mans Université
Spanish, English, French, German