About me

Director at Meta FAIR working on embodied AI, with a rare full-stack profile across frontier AI research, systems engineering, and organizational leadership. Previously founded MyoLab, and at Meta AI led the development and launch of foundational open-source infrastructure – FairScale and xFormers, (as TPM) and MyoSuite (as researcher and technical lead) – now totaling 15k+ GitHub stars and 125M+ downloads. Previously led 30+ person cross-functional teams at IBM Research.

I am also an Adjunct Lecturer at King’s College London.

Background

I previously founded MyoLab, developing next-generation embodied AI for human motor control. Before that, at Meta AI as a technical program manager (TPM) and researcher, I supported several large and impactful projects such as FairScale and xFormers, EgoExo and MyoSuite. Earlier, at IBM Research, I was a researcher, senior manager, and strategist leading a cross-functional team of 30+ developers, designers, and project managers to translate emerging research into prototypes, MVPs, and production systems, and co-leading the Global Technology Outlook (GTO).

My scientific background is in neuroscience and motor control. I hold a Ph.D. from the International Max Planck Research School at the University of Tübingen (summa cum laude), and conducted postdoctoral research at the Karolinska Institutet and MIT’s McGovern Institute for Brain Research. My work includes first-author publications in Nature, Science, Cell, NeurIPS, ICML, and ICRA, with a total citation impact factor > 300 (lead-author IF > 200).

Selected Work & Projects

A selection of projects, benchmarks, and community efforts across embodied AI, machine learning systems, and neuroscience.

2025

NeurIPS-MyoChallenge 25 -- Towards Human Athletic Intelligence Vittorio Caggiano, et al Website | code

2024

NeurIPS-MyoChallenge 24 -- Physiological Dexterity and Agility in Bionic Humans Vittorio Caggiano, et al Website | code

2023

	NeurIPS-MyoChallenge 23 -- Towards Human-Level Dexterity and Agility. Vittorio Caggiano, et al Website | code
	MyoDex: A Generalizable Prior for Dexterous Manipulation Vittorio Caggiano, Sudeep Dasari, Vikash Kumar Website | code
	SAR: Generalization of Physiological Dexterity via Synergistic Action Representation Cameron Berg, Vittorio Caggiano, Vikash Kumar Website | Publication | code
	MyoSuite Web Demo: An interactive demonstration of the Musculoskeletal Models of the MyoSuite framework. Vittorio Caggiano, Vikash Kumar MyoSuite Demo | code

2022

	NeurIPS-MyoChallenge: Learning contact-rich manipulation using a musculoskeletal hand. Vittorio Caggiano, Huawei Wang, Guillaume Durandau, Seungmoon Song, Yuval Tassa, Massimo Sartori, Vikash Kumar Challenge website | Twitter
	MyoSuite: An embodied AI platform that unifies neural and motor intelligence Vittorio Caggiano, Huawei Wang, Guillaume Durandau, Massimo Sartori, Vikash Kumar Meta Tech@ blog | code
	MyoSim: Fast and physiologically realistic MuJoCo models for musculoskeletal and exoskeletal studies Huawei Wang*, Vittorio Caggiano*, Guillaume Durandau, Massimo Sartori, Vikash Kumar code
	xFormers: A modular and hackable Transformer modelling library Benjamin Lefaudeux, Francisco Massa, Diana Liskovich, Wenhan Xiong, Vittorio Caggiano, Sean Naren, Min Xu, Jieru Hu, Marta Tintore, Susan Zhang code

2021

	Fully Sharded Data Parallel: faster AI training with fewer GPUs Myle Ott, Sam Shleifer, Min Xu, Priya Goyal, Quentin Duval, Vittorio Caggiano Facebook Engineering blog | code | docs
	FairScale: A general purpose modular PyTorch library for high performance and large scale training Mandeep Baines, Shruti Bhosale, Vittorio Caggiano, Naman Goyal, Siddharth Goyal, Myle Ott, Benjamin Lefaudeux, Vitaliy Liptchinsky, Mike Rabbat, Sam Sheiffer, Anjali Sridhar, Min Xu code