StellenbeschreibungppAmazon RIVR is a robotics company pioneering Physical AI through real-world doorstep delivery. Founded in 2024 as an ETH Zurich spin-off, RIVR developed wheeled-legged robots designed to operate in complex, unstructured environments such as stairs, gates, doors, and uneven urban terrain. We believe that achieving general physical intelligence requires solving real customer problems in the real world, where robots can learn from rich operational data at scale. /p pFollowing our acquisition by Amazon in March 2026, we are continuing this mission with greater reach and speed. By combining custom robot hardware, onboard autonomy, and cloud-based coordination, Amazon RIVR is building the next generation of safe, reliable autonomous robots for last‑mile delivery /p h3Job Description /h3 pOur global fleet of autonomous robots operates in the real world, generating vast amounts of multi‑modal sensor data. While our VLA team focuses on building large‑scale models to consume this data, much of it remains unlabeled and unstructured. We are seeking an expert in self‑supervised and representation learning to unlock the full potential of this massive data pool. /p pIn this role, you will be responsible for designing and building the core data engine that transforms raw, real‑world sensor data into high‑signal, structured datasets suitable for training neural networks. You will pioneer methods to automatically curate, filter, and pseudo‑label this data, creating powerful representations that serve as the foundation for all downstream tasks, including navigation, imitation learning, and decision‑making. /p pYou will work directly with the VLA and Reinforcement Learning teams to define data strategies and interfaces, ensuring the data you produce directly accelerates their model development. If you are passionate about solving the "data bottleneck" in robotics and want to build the systems that learn meaningful patterns from the physical world, we invite you to join us. /p ul liDesign, build, and maintain scalable data pipelines to process, filter, and transform terabytes of raw, multi‑modal sensor data (e.g., video, LiDAR, IMU, odometry) from our robotic fleet. /li liDevelop and implement state‑of‑the‑art self‑supervised and representation learning algorithms to automatically extract features, discover patterns, and generate pseudo‑labels from our unlabeled data. /li liCollaborate closely with the VLA Foundation Model and RL teams to define data requirements, APIs, and strategies for leveraging curated datasets and learned representations. /li liArchitect and implement robust evaluation strategies, benchmarks, and datasets to rigorously track the performance and quality of both the data pipeline and the downstream models that consume it. /li liOwn the data integration workflow, creating efficient data loaders and access patterns to make high‑signal data readily available for model training and experimentation. /li liResearch and prototype novel techniques in data curation, active learning, and anomaly detection to continuously improve the quality and efficiency of our data engine. /li liMaster’s degree or higher in a relevant field such as Computer Science, Machine Learning, or Robotics. /li liA minimum of three years of industry or research experience, with PhD experience applicable. /li liDeep expertise in self‑supervised learning (SSL) and representation learning, particularly with multi‑modal sensor data (e.g., contrastive learning, masked autoencoders, world models). /li liProven experience in building and managing large‑scale data processing pipelines for machine learning (e.g., using Spark, Kubeflow, or similar cloud‑native tools). /li liStrong understanding of robotic sensor data (e.g., camera, LiDAR, IMU, odometry) and their characteristics. /li liStrong programming skills in Python and deep experience with PyTorch, including creating custom and efficient DataLoaders. /li liExperience with MLOps best practices and data versioning tools (e.g., DVC, Pachyderm) /li liPhD degree in Robotics, Engineering, Computer Science, Machine Learning or a similar discipline, or an equivalent amount of research experience. /li liPublications at top‑tier ML or robotics conferences (e.g., NeurIPS, ICML, CVPR, CoRL, ICLR). /li liExperience with generative models (e.g., GANs, Diffusion Models) for data augmentation or simulation. /li /ul pAmazon RIVR is committed to building a diverse and inclusive team that values every perspective. If you’re passionate about driving innovation in robotics and creating meaningful impact, we encourage you to apply and bring your unique self to our team. /p pWe believe the best work is done when collaborating and therefore require in‑person presence in our office locations. /p /p #J-18808-Ljbffr
