VLA Research Scientist

Company Description We are a team of roboticists, AI engineers, and AI scientists with roots in Harvard and MIT, dedicated to advancing vision-action-language (VLA) foundational AI models. Our work focuses on building the data layer that powers intelligent, moving agents, such as humanoid robots and other autonomous systems. We collaborate across disciplines to design, curate, and leverage large-scale multimodal datasets that link perception, decision-making, and action. The organization values rigorous research, practical impact, and a culture of innovation where team members can shape the future of AI-enabled robotics. Joining the team offers the opportunity to contribute to cutting-edge projects that bridge academic research and real-world applications.Role Description The VLA Research Scientist will conduct end-to-end research on vision-action-language models, including problem formulation, model design, experimentation, and evaluation. This full-time role is on-site in Paris and involves developing new architectures and training strategies for multimodal AI systems that control robotic agents. Day-to-day responsibilities include designing and running experiments, analyzing large-scale datasets, collaborating with engineers on data collection and simulation pipelines, and publishing results in leading conferences and journals when applicable. The role also includes prototyping and validating models in real or simulated robotic environments, reviewing emerging literature, and contributing to shared tools and frameworks used across the team. The scientist will work closely with other researchers and stakeholders to turn research insights into robust, deployable systems.Qualifications Strong expertise in machine learning and deep learning, including multimodal models, representation learning, and sequence or transformer architectures.Proficiency in programming and research tooling, such as Python, PyTorch or TensorFlow, version control, and experiment management frameworks.Background in robotics or embodied AI, including working with robot control, simulation environments (e.g., MuJoCo, Isaac, or similar), or reinforcement/behavior learning.Experience with data-centric AI, including dataset design, collection, cleaning, annotation, and evaluation for vision and language tasks.Demonstrated research experience, such as prior publications, technical reports, or substantial projects in AI, robotics, or related fields.Advanced degree in Computer Science, Robotics, Electrical Engineering, or a related discipline (PhD preferred; strong MSc with research experience considered).Ability to design rigorous experiments, interpret results, and communicate findings clearly in written and spoken English.Collaborative mindset, willingness to work on-site in Paris, and ability to operate in a fast-paced, highly innovative research environment.