Grzegorz Bartyzel
Applied Scientist · Machine Learning & Robotics
Kraków, Poland
Hi there! Looking back at my career trajectory, I would call myself an Applied Scientist, or perhaps an Applied Research Engineer, with a specialization in robotics and AI.
I’ve been developing autonomous systems for over 9 years, and along the way, I also completed a PhD at the AGH University of Krakow (you can read my doctoral thesis on this page). Because of my PhD, my scientific interests lie in reinforcement learning and representation learning—combined, those fields can yield some great results ;).
During my professional journey, I have worked with a broad range of autonomous systems, from industrial robots to autonomous vehicles, giving me a very comprehensive understanding of the field. It’s also worth mentioning that I’m not just a research guy, as my past projects have also involved some pretty heavy software engineering ;)
9+
Years Experience
4
Patents
5
Publications
Experience
Founding Research Engineer - Contact-Rich Manipulation
Staff Data Scientist - Physical AI
Led the company's advanced policy control initiative, integrating VLA, ACT, and RL models into the robotic platform.
AI Staff Engineer - AI Planning
Led AI trajectory planning for highway/urban and a classic path-planning system for L2+ off-road convoying.
Robotics Software Engineer
Designed LiDAR-based detection and path-planning algorithms.
AI Specialist
Designed a deep RL robotic insertion system achieving over 95% success rate.
Junior Electronics Engineer (Internship)
Designed, prototyped, and validated an electronic control unit for electric arc detection in mild-hybrid vehicles.
Selected Publications
Read my Ph.D. Thesis (PDF)IEEE Robotics and Automation Letters (RA-L), 2024
Multimodal Variational DeepMDP: An Efficient Approach for Industrial Assembly in High-mix, Low-volume Production
Proposed a method combining multimodal variational autoencoders and deep RL to improve generalization in contact-rich robotic manipulation tasks.
Journal of Intelligent & Robotic Systems, 2023
Reinforcement learning with stereo-view observation for robust electronic component robotic insertion
Developed a robust deep RL approach leveraging stereo vision to successfully solve high-precision robotic insertion tasks.
IEEE Intelligent Vehicles Symposium (IV), 2024
MIMP: Modular and Interpretable Motion Planning Framework for Safe Autonomous Driving
Co-authored a modular framework designed to handle complex, real-world autonomous driving scenarios safely and interpretably.
