Research Intern

About Trace Trace is building the data, infrastructure, and research that enable robots to learn from the real world. General-purpose robotics requires a deeper understanding of how robots learn from real-world data. Our research explores how to transform human experiences, including egocentric video, demonstrations, hand-object interactions, and multimodal sensory signals, into transferable representations, actionable knowledge, and generalizable robot behaviors. At our Brooklyn research lab, we combine fundamental research in robot learning, computer vision, and embodied AI with large-scale datasets, machine learning infrastructure, and real-world robotic experimentation. Our goal is to advance the science of robot learning while developing the datasets and learning foundations for increasingly capable, adaptable, and reliable robots. About the role We are looking for highly motivated Research Interns to join our research team and work on challenging, high-impact problems at the intersection of robot learning, embodied AI, and learning from human experience. This internship is designed for PhD students and exceptional Master's students who are passionate about advancing the state of the art in robotics and machine learning. As a Research Intern, you will take ownership of a focused research project, working closely with our scientists and engineers to formulate research questions, develop novel methods, design rigorous experiments, and evaluate results using real-world datasets, simulation, and/or robotic platforms. We encourage ambitious, scientifically grounded research with the potential to contribute to top-tier robotics, machine learning, and computer vision conferences. You will have opportunities to pursue original research, collaborate with an interdisciplinary team, and help shape Trace's long-term research directions. Research areas Research projects may span one or more of the following areas, but are not limited to: Learning from Human Experience and Egocentric Data Human-to-Robot and Cross-Embodiment Transfer Foundation Models for Robot Learning Visual-Tactile and Multimodal Robot Learning Data-Centric Robot Learning and Scaling Dexterous Manipulation and Generalizable Robot Skills Robustness, Generalization, and Reliable Robot Learning We particularly welcome candidates who bring novel perspectives, strong methodological expertise, or creative research ideas that complement these directions. What you'll do Lead an original research project from problem formulation and hypothesis development through method design, implementation, experimentation, and evaluation, with guidance and collaboration from Trace's research team. Develop and evaluate novel robot learning methods, including learning algorithms, representations, models, or training strategies that leverage human demonstrations, multimodal datasets, and robotic interaction data. Design rigorous experiments and benchmarks to investigate fundamental questions about data quality, human-to-robot transfer, policy learning, generalization, and real-world robotic performance. Work with diverse data modalities, potentially including egocentric video, human hand and body motion, teleoperation demonstrations, visual-tactile signals, force measurements, and robot trajectories. Collaborate with research scientists and engineers to build reproducible experimental pipelines, train and evaluate models, and validate research findings in simulation and/or on real robotic platforms. Analyze experimental results and failure modes, draw actionable scientific insights, and communicate findings through technical discussions, presentations, and written reports. Contribute to high-quality research publications and open research artifacts, with opportunities to submit work to leading robotics and machine learning venues. What we're looking for Currently pursuing a PhD or Master's degree in Robotics, Machine Learning, Computer Science, Computer Engineering, Electrical Engineering, Mechanical Engineering, or a closely related field. Strong research background in at least one relevant area, such as robot learning, imitation learning, reinforcement learning, computer vision, embodied AI, foundation models, dexterous manipulation, multimodal learning, or learning from human demonstrations. Demonstrated research excellence and potential, evidenced by high-quality publications, impactful research projects, open-source contributions, or other substantial research accomplishments. Publications at leading venues such as RSS, CoRL, ICRA, IROS, T-RO, IJRR, NeurIPS, ICML, ICLR, or CVPR are highly valued. First-author or co-first-author publications are a strong plus. Excellent programming skills in Python and hands-on experience with modern deep learning frameworks such as PyTorch or JAX. Experience with C++ and parallel programming, such as CUDA, is a plus. Strong communication, collaboration, and problem-solving skills, with the ability to work effectively in a fast-paced, interdisciplinary research environment. Nice to have Experience with real-world robot learning experiments, robotic manipulation platforms, or deployment of learned policies on physical robots. Experience with egocentric video, human demonstrations, hand tracking, pose estimation, motion retargeting, or cross-embodiment learning. Experience with vision-language-action models, diffusion-based policies, imitation learning, reinforcement learning, or world models. Experience with tactile sensing, force estimation, multimodal sensor fusion, or contact-rich manipulation. Experience with robotics simulation environments and tools such as MuJoCo, Isaac Sim, or related frameworks. Previous research experience in an industrial research lab or a collaborative, interdisciplinary research environment. Trace is an equal opportunity employer. #J-18808-Ljbffr Trace