Member of Technical Staff, Microsoft Robotics (Spatial AI)

Design, develop, and evaluate physical world models that capture 3D spatial structure, object geometry and pose, physics dynamics, material properties, and semantic scene understanding for robotic applications. Build and train world models (e.g., video prediction models, neural physics simulators, 3D generative models, scene graph representations) that predict future states of physical environments conditioned on robot actions, enabling model-based planning and policy learning. Develop spatial AI capabilities including 3D scene reconstruction, object detection and pose estimation, spatial relationship reasoning, occupancy prediction, and dense 3D feature representations for robot perception and planning. Implement and maintain evaluation frameworks for world models and spatial AI systems, including prediction accuracy metrics, planning performance benchmarks, and generalization testing across environments and object categories. Collaborate with robotics researchers, learning engineers, and simulation engineers to integrate world models into robot planning and control pipelines, enabling model-predictive control, imagination-based planning, and data-augmented training. Build data pipelines for training world models, including multi-sensor data fusion (RGB, depth, LiDAR, proprioception), scene annotation, and dataset curation for diverse physical environments and interaction scenarios. Write efficient, readable, extensible code in Python (including PyTorch, JAX, or TensorFlow) for model development, training, and evaluation, leveraging GPU computing infrastructure for large-scale experiments. Contribute to the formulation of the team's world modeling research and development roadmap, identifying high-impact technical directions and collaborating with leadership to prioritize investments. Present research findings and model evaluation results clearly and efficiently to internal stakeholders and external partners, contributing to technical publications, blog posts, and conference presentations. Stay current with state-of-the-art research in world models, spatial AI, 3D vision, neural physics simulation, and foundation models for physical understanding, actively contributing to the body of thought leadership in these areas. Bachelor's Degree in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field AND 2+ years data-science experience (e.g., managing structured and unstructured data, applying statistical techniques and reporting results) OR Doctorate in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field OR Master's Degree in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field AND 1+ year(s) data-science experience (e.g., managing structured and unstructured data, applying statistical techniques and reporting results) or consulting experience Bachelor's Degree in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field AND 5+ years data-science experience (e.g., managing structured and unstructured data, applying statistical techniques and reporting results) OR Master's Degree in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field AND 3+ years data-science experience (e.g., managing structured and unstructured data, applying statistical techniques and reporting results) OR Doctorate in Data Science, Mathematics, Statistics, Econometrics, Economics, Operations Research, Computer Science, or related field AND 1+ year(s) data-science experience (e.g., managing structured and unstructured data, applying statistical techniques and reporting results) OR equivalent experience. Strong background in robotics perception, navigation, and proprioceptive sensor stacks and systems integration, including algorithm and model development and implementation in real-world applications. Experience with world models, video prediction models, neural physics simulators, or generative 3D models for physical environment understanding and prediction. Strong background in 3D computer vision, including depth estimation, 3D reconstruction, NeRF/Gaussian splatting, point cloud processing, or spatial reasoning. Proficiency in PyTorch, JAX, or TensorFlow with experience training large-scale models on GPU clusters (Azure Machine Learning, Kubernetes, or equivalent). Experience with robotics perception systems, including multi-sensor fusion (RGB-D, LiDAR, proprioception), object pose estimation, or scene graph construction. Familiarity with model-based reinforcement learning, model-predictive control, or imagination-based planning approaches that leverage learned world models. Published research or demonstrated contributions to world models, spatial AI, 3D vision, neural simulation, or physical AI in top-tier venues (NeurIPS, ICML, ICLR, CoRL, RSS, CVPR, ECCV, or equivalent).