Hi there 👋

My name is Kirill Kolodiaznyi and I'm a software developer.

🔬 Research & Intersest

The research I’m working on now is the implementation of a robotic platform based on the NVIDIA Jetson Nano 2Gb device with monocular SLAM navigation. The main aim of this work is to improve feature matching algorithms used for camera positioning in a SLAM pipeline. The recent experiments are about using deep learning for SLAM, especially using the Local Feature Matching with Transformers for camera positioning.

Another topic of my interest is multi-threaded algorithms and using them for performance improvements in projects I’m working on. The result of this work is the library that represents higher-level, task-based parallelism. It’s based on the idea of the Intel TBB work-stealing threads management approach.

Also, I’m interested in learning and applying the C++ language for solving Machine Learning tasks. I wrote several articles and a book about the application of C++ to ML. Additionally, I made C++ implementations of popular object detection methods like Faster R-CNN and Mask R-CNN.

✍ Publications

• Hands-On Machine Learning with C++: Build, train, and deploy end-to-end machine learning and deep learning pipelines

• Local Feature Matching with Transformers for low-end devices Feb 1, 2022, arXiv:2202.00770 [cs.CV]

🔧 Technologies & Tools