Ruslan Murtazin

- Led R&D of Russian Sign Language (RSL). - Successfully trained a transformer for recognition of RSL reaching a 75% recognition rate. - Implemented optimizations of a model using ONNX that significantly accelerated model inference by 10 times. - Leveraged advanced data visualization and analysis data using Pandas and Plotly to estimate model predictions. - Top-3 position on Paperswithcode: https://paperswithcode.com/paper/fine-tuning-of-sign-language-recognition

- Led the process of development and training of pose estimation models - training, and optimization of hardware models. - Accelerated the speed of pose estimation models by 3 times using TensorRT technology, which made it possible to reduce equipment costs in industrial security projects and ensure the safety of people

- Led the training and implementation of the YOLO model for in-line inspection of gas and oil pipelines, demonstrating exceptional technical experience and industry knowledge. - Improved the efficiency of the detection model, which is 37 times lighter and 17 times faster than the previous model, resulting in significant cost savings and increased productivity for the company.