San FranciscoJune 11, 2026 /PRNewswire/ — Reka, an AI research lab focused on developing foundational intelligence technologies for the real world, announced a partnership with Moonvalley to bring in a group of outstanding AI researchers and engineers, jointly accelerating the advancement of embodied AI models and infrastructure.
Following this integration, Dr. Mateusz Malinowski, Dr. Mikołaj Bińkowski, along with a team of researchers and engineers who previously worked at DeepMind, Meta, Amazon, Microsoft, Google, Wayve, and Runway, have officially joined Reka. Malinowski, formerly a senior research scientist, and Bińkowski, previously a senior research scientist at Google DeepMind, were both core researchers on Google’s Veo series of models and will now join Reka’s research leadership team.
Reka is dedicated to building AI capable of reasoning, simulating, and acting in real-world environments. Its models are designed to tackle complex real-world challenges, with applications spanning autonomous defense, intelligent robotics, next-generation wearable devices, and media.
Dani Yogatama, CEO of Reka, stated: “Embodied AI is fundamentally a cutting-edge research challenge. This team possesses top-tier expertise in video generation and multimodal models. Together, we will build AI systems that can understand and act in the real world, covering areas such as simulation, robotics applications, and real-time decision-making.”
Mateusz Malinowski commented: “The models we develop are not just capable of generating videos; they can understand the operational rules of the real world. They can simulate motion states, physical laws, and temporal dynamics, allowing robots to predict outcomes before acting, creating safer and more responsible AI. Our approach of integrating simulation and reasoning into a single system is the right direction aligned with industry trends.”
The integrated team will focus on developing the World-Language-Action Model (WLAM). This is a unique, all-in-one model trained on first-person perspectives and various real-world data. It can perform planning through highly realistic simulations, thereby enabling perception and action in the real world. The model can generate high-fidelity multimodal content, leverage advanced spatiotemporal awareness to convert multimodal perceptual information into robotic execution commands, and perform fine-grained reasoning on long-form and streaming video.
Reka’s current operations are primarily divided into four segments:
- Reka Labs: A research and development platform for foundational intelligence technologies for the real world
- Reka Infer: A multimodal AI reasoning platform and application programming interface (API)
- Reka Vision: A video processing platform supporting annotation, reasoning, retrieval, and editing of massive video data
- Reka Claru: An embodied AI training data platform, providing large-scale first-person perspective videos, robot motion trajectories, world model imagery, and professional human judgment data
Reka is actively recruiting for research, engineering, and product roles. Enterprise partners seeking priority access to the Reka platform can directly contact the team for discussions.