Research Scientist: Pretraining

ML Engineer — Robot Foundation Model PretrainingWe are building large-scale embodied intelligence systems designed to operate in complex real-world environments. Our work focuses on training robot foundation models using massive multimodal datasets spanning video, language, proprioception, action traces, and related sensor modalities.We are seeking ML Engineers to design and execute large-scale pretraining efforts that develop general-purpose capabilities across embodiments, tasks, and environments. This role focuses on turning raw robotic interaction data into scalable, generalizable intelligence.What You’ll DoDesign and Execute Large-Scale PretrainingLead or contribute to large-scale training runs for robot foundation modelsWork with transformer-based and diffusion-based architectures for multimodal learningDefine model architectures, training objectives, and curricula for multimodal robotic dataDevelop Data and Training StrategiesDesign scalable data mixtures and sampling strategies across petabyte-scale datasetsStructure training curricula across vision, action, state, language, and other modalitiesConvert raw robotic interaction data into usable signals for model trainingImprove Model Performance Through AnalysisRun ablation studies to understand scaling laws, data quality effects, and architecture tradeoffsAnalyze training dynamics and large-model failure modesIterate on data and training design to improve generalization and robustnessCollaborate on Large-Scale SystemsWork closely with ML infrastructure and systems teams to improve cluster utilization, throughput, and reliabilityContribute to multi-node, multi-GPU distributed training effortsEnsure efficient execution of large-scale training workflowsShape Data Collection and Model DirectionGuide data collection efforts toward high-impact capabilities and gapsIdentify and integrate new datasets to improve model coverage and generalizationBridge raw data generation with downstream model performanceWhat We’re Looking ForExperience training large-scale transformer or diffusion models (e.g., language, video, audio, or generative models)Experience with multi-node, multi-GPU distributed training systemsStrong understanding of:Scaling lawsOptimization dynamicsLarge-model training behavior and failure modesStrong proficiency in PyTorch and ability to debug across the full ML stackComfort working with large-scale datasets and rapid experimentationStrong empirical rigor combined with fast iteration speedPreferred ExperienceExperience with multimodal or generative model training at scaleBackground in robotics, embodied AI, or sequential decision-making systemsExperience with large distributed training infrastructureFamiliarity with curriculum learning or dataset mixture designExperience analyzing large-scale training runs and extracting actionable insightsWhy This Role MattersBuild the core intelligence layer for general-purpose robotic systemsDirectly shape how robots learn from large-scale real-world dataWork at the intersection of large-scale AI training, systems, and roboticsDevelop foundation models that generalize across embodiments and environmentsAbout the CompanyWe are a research-driven AI and robotics company focused on building scalable embodied intelligence systems. By combining advances in machine learning, large-scale training infrastructure, and robotics, we aim to develop systems capable of operating robustly in the physical world.We are committed to building an inclusive and diverse workplace and encourage applicants from all backgrounds to apply.