Lofty aspirations require strong foundations. At Torc, the Infrastructure & Tools team is tasked with creating and maintaining the foundational systems on which the rest of the organization depends. The team is composed of cross-functional, multi-disciplinary software professionals who are dedicated to creating the infrastructure, web applications, and backend tools that enable testing and data analysis for our self-driving vehicles. Each plays a unique role in helping to realize Torc’s mission of making the world safer.
From the Road to the Cloud
You might not notice the Infrastructure & Tools team while making your way through Torc Robotics HQ. That’s because much of their work takes place behind the scenes. They provide the backbone for testing ops, working to solve the problems of scalability and reliability in the self-driving vehicle space. From on-vehicle troubleshooting to large-scale cloud deployments, they construct cutting-edge solutions that empower engineers to build, deploy, test and analyze autonomous software.
On-vehicle Computing
The Infrastructure & Tools team works closely with test engineers to ensure that the on-vehicle computing hardware meets performance requirements. They use industry-standard tools and services to manage our builds and deployments, including Git, Jenkins, Ansible, and Docker. They are actively involved in the design of next-generation hardware systems in the self-driving vehicles. It is up to them to ensure a stable on-vehicle computing platform to reliably support increasing on-road testing requirements.
Cloud Infrastructure
The team is responsible for provisioning, configuring and deploying simulation environments and log analysis pipelines to cloud compute and storage platforms using Kubernetes, terraform and other tools. Log analysis of petabyte-scale real-world vehicle test data and auto-scaling of simulation tasks requires innovative techniques to balance trade offs between cost and performance by creating ephemeral instances when cloud provider demand is low and on-demand instances for time-sensitive tasks.