Назым Агыбай

I'm building computer vision systems at Taharqa Technologies, where I develop visual navigation methods using image matching, neural networks, and camera motion estimation.

I optimize data-processing pipelines and streaming-video preprocessing for stable real-time performance, analyze navigation accuracy, and prepare neural network architectures for ONNX conversion while minimizing export-related accuracy loss.

Previously at KMG Engineering, part of KazMunayGas Group, I developed and fine-tuned a YOLOv8-Pose model for human pose estimation at industrial facilities. I annotated keypoints in CVAT, prepared training datasets, and used machine learning to support industrial safety monitoring.

My graduation work applied YOLOv8, Python, and OpenCV to analyze passenger boarding and alighting in public transport. I'm grounded in mathematical and computational sciences, with coursework spanning machine learning, optimization, statistics, big data analysis, and reinforcement learning.