Vasileios Lioutas
Machine learning researcher with 8+ years of experience advancing AI research and translating it into production systems, with applications spanning autonomous driving, NLP, and beyond. At Inverted AI, I lead the development of novel deep learning simulation architectures for realistic multi-agent driving behaviors, transitioning research directly into commercial products adopted by AV industry leaders. I have contributed to open-sourced TorchDriveSim, and published 20+ papers in premier venues including ICML, ICLR, NeurIPS, and ACL.
I received my Ph.D. in Computer Science from the University of British Columbia (2025), advised by Dr. Frank Wood, with a thesis on realistic controllable driving simulators. I hold an M.Sc. from Carleton University (GPA 12.00/12.00, Senate Medal for Outstanding Academic Achievement) and a B.Sc. from Aristotle University of Thessaloniki. I was a recipient of the NSERC CGS-D Doctoral Scholarship (2021–2024).
Prior to Inverted AI, I was an ML Research Associate at Huawei Noah's Ark Lab, working on model compression and multilingual neural machine translation, delivering 4–7× compressed models for production edge deployment and publishing 3 papers at ACL and EMNLP. I also built real-time recommendation systems as a Machine Learning Engineer at MEDIAFORCE.ca.
Email / Google Scholar / GitHub / LinkedIn / Resume (PDF)
Selected Publications
I'm broadly interested in generative modeling, probabilistic inference, and reinforcement learning, with a focus on building realistic and controllable simulations of complex real-world environments. My research has primarily centered on autonomous driving simulation, developing models that generate realistic multi-agent traffic behaviors and can be steered toward desired outcomes. Driven by a passion for solving hard technical problems at the intersection of research and real-world impact, I'm excited to push the boundaries of what AI systems can do next.
Full list on Google Scholar.
