We hire people who want to build the future of food without having to look at the past.
Because the problem is enormous, the mandate is real, and the results are already on the board. In fifteen months this team took the most dangerous job in American manufacturing to zero injuries — by taking it to zero people. If you want your work to show up in a number the whole industry has to respond to, this is the desk.
Own hard classification problems under occlusion, feathers, and motion blur, at 15,000 units an hour. You'll work at the frontier of production VLA systems with training runs that ship to nine facilities weekly. Strong background in vision transformers or VLA architectures required; industrial experience not required. Nobody here had any.
Shape how our models see the line. You'll design classification taxonomies, hunt hesitancy events to extinction, and keep production behavior aligned with operational values as the fleet scales. Experience with RLHF or fine-tuning pipelines strongly preferred.
Nine facilities, 16,000 cameras, 1.4 million units a day, one scheduler. You'll own the telemetry and orchestration backbone that keeps network-wide rebalancing under a second. Kubernetes at the edge, time-series at absurd scale, and uptime numbers you'll want framed.
Monitor four facilities from wherever you live. You'll watch dashboards, verify automated resolutions, and escalate anomalies — the calmest job this industry has ever offered. Many of our observers came from the line; processing experience is valued but not required. On-site: no. No one is on-site.
Help the world understand what we're building — and what it means for workers, for consumers, and for the future of American food. You'll own our story across press, policy, and public channels, and you'll have the best numbers in the industry to tell it with.
