Sr CVML Engineer, Autonomy

Sourced from remoteforge.2kool4u.net

Blue River Technology, part of John Deere, is hiring a Senior CVML Engineer to develop production computer vision models for an autonomous mower initiative. The role involves depth estimation, segmentation, and tracking, optimizing deep learning models for real-time inference on embedded NVIDIA GPUs, and integrating perception into the robotics stack. This is a full-time remote position within the US with visa sponsorship available.

• Design, train, and deploy production-grade deep learning models for depth estimation, segmentation, and tracking.
• Optimize models for real-time, sub-millisecond inference on embedded NVIDIA GPU platforms using TensorRT, mixed-precision, and quantization-aware training.
• Own model development end-to-end, from prototyping to deployment in the perception pipeline.
• Collaborate with the robotics team to integrate perception outputs and architect onboard pipelines maximizing GPU saturation across multiple camera feeds.
• Analyze site data, identify failure cases, and continuously improve model robustness against challenging outdoor conditions.
• Use real and synthetic data strategies to train models and validate perception pipeline performance.

• Bachelor's or Master's degree in Computer Science or a related field; graduate degree preferred.
• 5+ years of post-academic experience building and deploying production vision models on real-time embedded systems.
• Hands-on expertise with deep learning frameworks, especially PyTorch, and deep understanding of modern computer vision architectures.
• Strong proficiency in TensorRT, NVIDIA ModelOpt, and ONNX Runtime for advanced model optimization.
• Strong Python programming skills for model development and strong C++ for performance-critical code.

• Direct experience deploying real-time vision models on physical robots or autonomous vehicles.
• Demonstrated ability to build and deploy end-to-end ML pipelines (data ingestion, training, testing, deployment).
• Experience with on-device model optimizations using TensorRT, ONNX, or similar engines.
• Experience with hardware-accelerated compression techniques, knowledge distillation, or network structural pruning.
• Familiarity with robotics frameworks like ROS / ROS2.