Bio
Co-founder and CTO of Dyna Robotics, a robotics company building general-purpose robots powered by embodied AI foundation models. Previously a research scientist at DeepMind focused on foundation models for robotics.
Session (1)
Additional Speaker Details
Using publicly available information we constructed an analysis to help you get a feel for this speaker before deciding to attend their session.
Jason Ma is a researcher-turned-founder, co-founder of Dyna Robotics, which built DYNA-1 — a commercial-ready robot foundation model reported to achieve a 99%+ autonomous success rate in continuous 24-hour operation. As lead author of Eureka and DrEureka, his work on LLM-guided reward design and sim-to-real transfer is directly relevant to attendees interested in moving robot learning from the lab into production systems.
Speaking style
Recent talks (2)
GitHub
@JasonMa2016- JasonMa2016/GoFAR★ 38Official repository for Paper 'Offline Goal-Conditioned Reinforcement Learning via f-Advantage Regression' (NeurIPS 2022)
- JasonMa2016/SMODICE★ 30Official repository for paper 'Versatile Offline Imitation from Observations and Examples via Regularized State-Occupancy Matching' (ICML 2022)
- eureka-research/Eurekaorg★ 3,200Official repository for 'Eureka: Human-Level Reward Design via Coding Large Language Models' (ICLR 2024); Yecheng Jason Ma is first author
- eureka-research/DrEurekaorg★ 938Official repository for 'DrEureka: Language Model Guided Sim-To-Real Transfer' (RSS 2024); Yecheng Jason Ma is co-first author
Recent writing (2)
- Dyna Robotics Unveils DYNA-1: The First Commercial-Ready Robot Foundation Model · other · Apr 2025
- Dyna Robotics Raises $120 Million to Advance Robotic Foundation Models · other · Sep 2025
Podcasts & interviews (1)
- He's Building $600M Robots That Fold Your Laundry · Founders in Motion with Thea Ngo · Sep 2025