Source description
About the role
About Corvus
Every physical good spends time in a warehouse, and every warehouse tracks their inventory. Today, nearly 100% of warehouses track their inventory manually using barcode scanners and climbing forklifts.
We're Corvus Robotics https://www.corvus-robotics.com/. Our fully autonomous Corvus One™ https://blog.corvus-robotics.com/corvus-one-launch-and-series-a-funding drones use computer vision & robotics to automatically track inventory, improving worker safety and increasing labor efficiency. We believe that data-driven, safe inventory management will optimize the global physical economy and improve economic prosperity for humanity.
About THE Role
With a growing fleet of autonomous drones and an expanding customer base, we're now ready to multiply ML iteration speed and unblock more advanced ML product delivery.
We're hiring a systems-oriented Senior Software Engineer to build the data infrastructure, training pipelines, and internal tooling that our ML team needs to move faster.
Specifically in this role you will:
-
Build and maintain the data pipeline infrastructure that consolidates internal infra, labeling tools, S3, and other data sources into a unified, queryable system
-
Build tooling for dataset selection and curation that can programmatically target specific data (by environment, object type, etc.)
-
Own ML data infra from robot to training run, accessible to the ML team without backend engineering help
-
Build model evaluation and regression testing infrastructure -- real metrics, not vibes or "someone complained in prod"
-
Automate the model retuning loop for standard tasks so ML engineers can be mostly hands-off on routine updates
This is a hybrid or remote role with periodic trips to HQ in Mountain View, CA.
MUST HAVES
-
2-3 years shipping real production ML infrastructure for big datasets, not just scripts
-
Experience building distributed data pipelines that consolidate multiple sources
-
Demonstrated understanding of data flow from raw collection, labeled training set, to trained models
-
Experience building systems from scratch, or contributed heavily to a small-team infra build where the playbook didn't exist
-
Ability to thrive in a startup environment with high ambiguity. You'll figure out what to build
NICE TO HAVES
-
Experience setting up annotation tooling and workflows
-
Background in robotics autonomy and computer vision
Experience integrating with tools like Kubeflow, SLURM, or similar for scalable training workflows
More at Corvus Robotics