Senior AI Engineer Self-Supervised Learning

Amazon RIVR seeks a Senior AI Engineer in Zürich to develop self-supervised learning systems for autonomous robot sensor data. The role focuses on building scalable data pipelines and representation learning models to advance Physical AI for doorstep delivery.

Amazon RIVR is advancing Physical AI through real-world autonomous delivery robots. This position involves creating large-scale models that process vast amounts of multi-modal sensor data from the global robot fleet. The primary goal is to overcome data bottlenecks in robotics by developing systems that learn meaningful patterns from physical environments.

Responsibilities

• Construct and upkeep scalable data pipelines to handle, filter, and convert massive volumes of raw multi-modal sensor data collected by the robot fleet.
• Create and apply cutting-edge self-supervised and representation learning methods to automatically identify features, uncover patterns, and produce pseudo-labels from unlabeled datasets.
• Partner with the VLA Foundation Model and reinforcement learning groups to establish data needs, APIs, and approaches for utilizing curated datasets and learned representations.
• Design and execute strong evaluation frameworks, benchmarks, and datasets to monitor the performance and quality of data pipelines and downstream models.
• Manage the data integration process by developing efficient data loaders and access patterns to ensure high-signal data is easily accessible for model training and experimentation.
• Investigate and prototype new methods in data curation, active learning, and anomaly detection to enhance the quality and efficiency of the data engine.

Requirements

• Master’s degree or higher in Computer Science, Machine Learning, Robotics, or a related discipline.
• At least three years of industry or research experience, with PhD experience considered applicable.
• Extensive knowledge of self-supervised learning and representation learning, especially with multi-modal sensor data.
• Demonstrated ability to build and manage large-scale data processing pipelines for machine learning.
• Solid understanding of robotic sensor data and its characteristics.
• Proficient programming skills in Python and extensive experience with PyTorch, including creating custom and efficient DataLoaders.
• Familiarity with MLOps best practices and data versioning tools.

Nice to have

• PhD in Robotics, Engineering, Computer Science, Machine Learning, or a related field, or equivalent research experience.
• Publications at top-tier ML or robotics conferences.
• Experience with generative models for data augmentation or simulation.

About the company

Amazon RIVR is dedicated to fostering a diverse and inclusive team that values every perspective. The company emphasizes collaboration and requires in-person presence in office locations to achieve the best results.

• Committed to building a diverse and inclusive team that values every perspective.
• Believes the best work is done when collaborating and requires in-person presence in office locations.