Machine Learning Engineer - Vision Foundation Models

Description

The company's Perception team builds how its purpose-built robotaxi understands its surroundings, including the cabin. This role works on vision models that read passengers, objects, and behavior inside the vehicle, supporting rider safety and day-to-day ride operations.

The company is building a ground-up, fully autonomous vehicle fleet and the ecosystem needed to launch it, aiming at mobility-as-a-service in cities. The team is fast-moving and focused on execution.

You would own a model area from start to finish, covering data, training, evaluation, and deployment under real-time and compute limits, and work with colleagues across the company. Your output affects rider safety and experience as the robotaxi service grows.

Responsibilities

• Design, train, and evaluate vision models that recognize passengers, objects, and behavior in the cabin, covering detection, pose, and action recognition.
• Take ownership of a model or feature area end to end: curate the data, write labeling guidelines, train and evaluate, and carry it through to deployment.
• Set the quality standard for your models through rigorous evaluation and error analysis, so they are dependable in live operation.
• Deploy models that fit within the real-time and compute budget shared with the rest of the onboard perception system.
• Work with product, operations, safety, and vehicle software teams to turn requirements into model specs, metrics, and validation plans.

Requirements

• MS or PhD in computer science, machine learning, computer vision, or a related field, plus 6+ years of industry experience (3+ years if you hold a PhD).
• Deep knowledge of computer vision and deep learning for image and video, including detection, pose and action recognition, and fine-grained classification.
• Hands-on work with modern architectures: convolutional and vision-transformer detectors, and self-supervised or foundation-model backbones such as DINOv3.
• Strong Python and PyTorch skills for designing, training, and evaluating models.
• Experience with vision-language models or foundation-model backbones, and with auto-labeling or synthetic data for long-tail and rare classes.
• Proven record of running the full ML lifecycle on real-world data: dataset curation, labeling guidelines, training, metric design (precision, recall, and false-positive rate at an operating point), and error analysis.
• Bonus: experience with in-cabin or occupant monitoring systems for drivers or occupants, ideally with multi-camera or fisheye imagery and low-light conditions.
• Bonus: experience deploying deep learning models to real-time, resource-constrained systems such as robotics, automotive, or edge devices, using TensorRT, ONNX, or similar.
• Bonus: proficiency in C++ for model integration.