Remote Lead Edge AI & Computer Vision Architect

Micasa Global; United States; On-site; USD 180,000 - 240,000; Full time

Job summary

Micasa Global is seeking a Lead Edge AI & Computer Vision Engineer to architect and optimize end-to-end CV and low-latency execution pipelines on NVIDIA Jetson embedded hardware. You will work on real-time detection, spatial math, and memory architectures to achieve millisecond-scale latency for edge deployments.

The role demands deep expertise in C++/CUDA, TensorRT, OpenCV, PyTorch, and a proven track record with Jetson ecosystems, PTQ/QAT, and Nsight profiling.

Qualifications

• 12-15+ years of experience bridging real-time CV algorithm design and low-level C++/CUDA on embedded edge hardware.
• Deep expertise with NVIDIA Jetson ecosystem (Orin NX, AGX Orin, L4T) and TensorRT FP16/INT8 engines.
• Expertise in optical/spatial geometry, camera calibration, 2D/3D transformations, and millimeter-level sizing in industrial settings.

Responsibilities

• Evaluate real-time detection topologies (e.g. YOLO) at 100+ FPS and implement 2D homography and camera calibration with tight error bounds.
• Implement post-training quantization and compile PyTorch/ONNX models into TensorRT engines for edge devices.
• Design zero-copy memory pipelines and DMA-based data transfers to minimize latency.
• Develop C++17 execution frameworks with multi-threading, lock-free structures, and asynchronous I/O.
• Instrument latency and roadmaps using Nsight Systems/NVTX to bound tail latency.

Skills

Edge AI

Computer Vision

C++/CUDA

TensorRT

Nsight Systems

Tools

Jetson Orin NX/AGX Orin

OpenCV

PyTorch

ONNX

CUDA

Job description

Micasa Global is seeking a Lead Edge AI & Computer Vision Engineer to architect and optimize end-to-end CV and low-latency execution pipelines on NVIDIA Jetson embedded hardware. You will work on real-time detection, spatial math, and memory architectures to achieve millisecond-scale latency for edge deployments.

The role demands deep expertise in C++/CUDA, TensorRT, OpenCV, PyTorch, and a proven track record with Jetson ecosystems, PTQ/QAT, and Nsight profiling.