Embedded Perception Engineer

About this role

Havoc seeks an Embedded Perception Engineer to architect and deploy real-time perception pipelines for autonomous maritime systems. You'll own vision models, sensor fusion, and edge AI optimization to enable autonomous vessels to sense and understand complex operational environments reliably.

What you'll do

• Design, train, and optimize vision models for object detection, classification, and tracking in maritime settings
• Develop end-to-end perception pipelines and deploy them to embedded edge hardware using Triton and similar tools
• Fuse multi-modal sensor data (camera, radar, lidar) into unified perception outputs and integrate with autonomy stacks
• Optimize model inference for latency and throughput through quantization, pruning, and hardware-specific tuning
• Monitor deployed system performance, investigate perception failures, and collaborate with teams to resolve field issues
• Partner with operators and customers to translate operational challenges into engineering improvements

What they're looking for

• Computer vision and deep learning (PyTorch, TensorFlow)
• Large vision models (ViT, CLIP, SAM)
• C++ and Python development on Linux
• NVIDIA Triton, TensorRT, ONNX Runtime
• Sensor fusion across heterogeneous modalities
• Embedded systems and edge inference optimization
• ROS2 or autonomy frameworks (preferred)
• Systems thinking and complex codebase navigation