About the role
The Role
We’re looking for a Robotics Data Infrastructure Engineer to own and build the data systems that power Verne’s robots in the real world. This is a hands-on founding engineer role with true ownership and freedom — your work will directly impact robots performing customer-critical tasks every day.
You will architect and deploy data pipelines on both AWS and edge devices, manage large-scale multi-modal datasets (images, video, time-series, text, etc.), and build the tooling that connects real-world robot data to training and evaluation workflows. You’ll work across the full robotics software stack, from ingesting sensor data and telemetry, to enabling large-scale policy learning pipelines that drive production robots.
Beyond writing great code, you’ll help drive technical decisions, lead cross-functional efforts, and bridge robotics, machine learning, and product requirements into scalable, reliable systems.
What You’ll Do
- Build and own our data backbone on AWS: Design and run cloud + edge pipelines using services like IoT Core, S3, ECR, Batch, ECS/EKS, and Step Functions. Your work keeps robot data flowing reliably and cost-efficiently from the field into the lab.
- Develop on-device data systems: Build robust, fault-tolerant data capture on edge PCs using MCAP/Protobuf, with clean schema contracts, buffering, and resumable uploads to the cloud.
- Wrangle massive multimodal datasets: Organize and version millions of images, videos, time-series (robot state, force/torque), and annotations. Enforce metadata, retention, and access patterns that scale.
- Build MLOps and DataOps pipelines: Automate data validation, labeling, augmentation, and model training/evaluation using containerized jobs and orchestrators like Batch, Step Functions, Airflow, or Prefect.
