Machine Learning Engineer
Summary
ML engineer at Herd-i, a Christchurch agritech company that films cows in the milking shed and uses computer vision to score lameness and body condition for farmers. Day to day you'd improve cow detection and keypoint models, build the MLOps pipeline, and ship DairyNZ-scale scores using Python, PyTorch/TensorFlow, YOLO/RT-DETR, XGBoost and AWS.
Herd-i | ML Engineer
Herd-i is a Christchurch agritech company using computer vision to score dairy cows for lameness and body condition. A camera sits in the shed and films each cow as she leaves the milking. The system identifies her by EID at the drafting gate, and the footage goes to the cloud, where ML models score her against the DairyNZ scale. Farmers see the results on a dashboard, with alerts and a watchlist. The point is to catch lame cows early, before they're obvious, so treatment is simpler and herd performance holds up. It's real, deployed ML on real farms, in NZ and the US.
The role is an ML Engineer on the core AI team, working under the Lead ML Engineer. You'd improve the cow detection and keypoint models, build out the MLOps pipeline, and work directly with the CTO and developers. It's a small team, so you'd own things rather than sit in a queue.
Tech stack (from the JD)
Language and frameworks: Python, PyTorch or TensorFlow
Detection: YOLO and RT-DETR for cow detection
Pose: keypoint detection, which drives the gait analysis
Scoring: regression models such as XGBoost
MLOps: CI/CD, containerisation, cloud (William said AWS is the core tech, so confirm that applies to the ML team)
Output: DairyNZ lameness scores (0–3) and body condition scores (3–6) feeding a web dashboard
AWS