People will come to you with solutions. Your job is to find the problem.
My manager taught me this, and it's the thing I come back to most. When someone pitches a fix, I listen, ask a lot of questions, and work back to what's actually hurting. If you skip problem discovery, you ship something nobody adopts.
Two sides of the autonomous vehicle data business: labeling the data, and collecting it in the first place.
Web apps used by clients, project managers and annotators
Where clients and project managers set up and run annotation projects. Teams build customized workflows for different kinds of annotation work, then track execution from start to delivery.
One tool for every common raw data and annotation type in AV and computer vision. It renders 3D LiDAR point clouds and sensor fusion tasks, combines 2D and 3D in a single task, and includes a QC tool and ML-assisted features.
Tools for running AV data collection fleets
Used by drivers in the field to track routes as they drive them, pick up assigned tasks, and keep the fleet coordinated.
Where clients and internal project managers create and manage projects, assign routes and tasks, and see progress as it happens.
Plans routes built around an operational design domain (ODD), so collection captures the variety AV models need. Routes then get assigned to drivers through the console.
I've also led multiple ML projects across the business, both for field data collection and for annotation.