Perception Engineer

About this role

Aurelius Systems, a VC-backed defense tech startup, seeks a Perception Engineer to develop and deploy computer vision and sensor fusion systems for autonomous directed energy (laser) counter-UAS platforms. You'll own the full perception stack—from model training and multi-sensor integration to real-time edge inference and field deployment—working with a small team that values shipping robust systems over research-only work.

What you'll do

• Design, train, and deploy ML/DL models (YOLO, RT-DETR, CNNs) for object detection, classification, and segmentation
• Integrate and fuse multi-modal sensor data (RGB, thermal, LiDAR/ToF, IMU) for real-time detection and tracking
• Develop image-processing techniques for low-SNR detection, deconvolution, and motion isolation
• Collaborate with hardware teams to integrate sensors (cameras, thermal imagers, LiDAR) and debug embedded firmware on microcontrollers and FPGAs
• Build data pipelines for ingestion, labeling, and augmentation of 100k+ labeled frames; optimize inference for ≥500 Hz edge throughput
• Create monitoring dashboards for model drift, hardware health, and automated retraining; mentor junior engineers

What they're looking for

• Computer vision and real-time object detection/tracking
• C++ with ML engineering experience
• ML frameworks (TensorFlow, PyTorch) and inference engines (TensorRT, OpenVINO)
• Sensor fusion and multi-sensor calibration
• Embedded systems (C/C++, Rust, VHDL/Verilog for microcontrollers and FPGAs)
• ROS2, Docker, and CI/CD for ML pipelines
• Edge-AI optimization (quantization, pruning)
• FPGA and embedded GPU platforms

Benefits

• Work on cutting-edge defense technology with real-world field testing on private 400-acre range
• Small, high-output team with unreasonable ambitions and minimal bureaucracy
• Access to labs in San Francisco and manufacturing hub in Detroit
• Collaborate with former US military operators and subject matter experts
• Opportunity to ship perception systems that operate against real targets in field conditions