Software Engineer - Planner Frameworks Pipeline

Zoox is looking for an experienced software engineer to work on large-scale simulation pipelines used to validate the behavior of the Zoox self-driving vehicle. These are data and GPU intensive workloads built on and Kubernetes. Given the massive scale and criticality of these pipelines, ensuring their reliability and efficiency has a significant impact on the company's ability to safely and quickly iterate on autonomy development. We are a small, scrappy team within the larger Autonomy organization. Although this role primarily involves off-vehicle pipelines, you will work closely with engineers developing the on-vehicle algorithms and models in our autonomy stack. We stay close to the end users - autonomy engineers - and think about the end to end use case for these validation pipelines. This is a hands-on role with a high degree of independence and ownership. You will be expected to contribute towards the framework’s architecture, reliability, efficiency, and grow its capabilities to support new use cases. You should have a track record of keeping production systems running with high availability. Experience with robotics or autonomous systems is not required but an understanding of the robotic data lifecycle is preferred. ### In this role, you will: - Improve the cost efficiency, reliability, and performance of our validation and simulation pipelines - Create production-grade APIs, SDKs, and tools to enable a varied set of validations of autonomous behaviors - Improve the ML training pipelines supporting the autonomous behavior org ### Qualifications - Bachelor’s degree in Computer Science or related field and 6+ years of industry experience - Experience optimizing large-scale distributed systems for cost and efficiency - Experience with AWS or similar providers - Proficiency with Python and familiarity with C++ ### Bonus Qualifications - Exposure to machine learning workloads (training, inference, data generation) from a cost optimization perspective - Background in algorithmic optimization or performance investigation - Experience with , particularly Ray Core and Ray Data