Computer Vision / Edge AI Engineer
I work on real-time computer vision, embedded inference, and perception systems, with a focus on taking algorithms from prototyping to resource-constrained hardware.
My main areas of interest are object detection and tracking, stereo vision, camera calibration, classical computer vision, and neural network deployment on edge devices.
Most of my commercial and production work is private or NDA-protected. The repositories below contain independent tools, experiments, and engineering projects related to the areas I work in.
YOLO → RKNN Converter Toolkit for deploying YOLO-family detectors on Rockchip NPUs. Covers ONNX graph modification, INT8 post-training quantization, RKNN conversion, and reference inference.
Stereo Camera Calibration Configurable OpenCV pipeline for intrinsic and stereo calibration, reprojection-error analysis, outlier filtering, and estimation of relative camera pose.
BrightSpotDetector Lightweight C++ / OpenCV detector for bright targets in grayscale, thermal, and IR imagery, with configurable ROI and detection parameters.
Electronics Workshop Personal hardware projects covering the complete prototyping cycle from circuit design and firmware to assembly and enclosure.
Computer Vision & ML OpenCV · PyTorch · YOLO · ONNX · RKNN · TensorRT
Programming Python · C++ · Go · Java · TypeScript · SQL
Edge & Systems Rockchip RK3588 · Raspberry Pi · ESP32 · Linux · Docker · MQTT · PostgreSQL