Mikhail Petrushkov

Experience

• 

  Itseez

  Jul 2015 - Nov 2015

  Software Engineer

  • Developed and optimized OpenVX extensions, resulting in significant speedimprovements of computer vision algorithms through efficient image processing oncustomer hardware.

• 

  Intel Corporation

  Feb 2017 - Oct 2018

  Software Engineer

  • Collaborated on the development, validation and bug fixing of an Intel cross-platformlibrary for developing media applications on Windows and Embedded Linux(MediaSDK), specifically for video encoding, decoding and video/image processing.• Successfully resolved critical customer issues.• Provided internal training to empower team members with the necessary knowledge.

• 

  IntelliVision

  Oct 2018 - Dec 2019

  R&D Engineer

  • Explored and implemented deep learning methods for object detection andclassification, including datasets preparation, model training and evaluation.• Optimized deep learning models for integration on embedded hardware, focusing ongraph optimization and model quantization.

• 

  Luxoft

  Jan 2020 - Jan 2021

  Software Engineer

  • Developed deep learning algorithms for efficient post-processing in the renderingpipeline.• Evaluated various methods for denoising Monte Carlo rendering results.• Contributed to the development of a cross-platform inference engine forGPU-accelerated inference of neural network models.

• 

  Intel Corporation

  Jan 2021 - now

  • Conducted research on deep learning algorithms for real-time image enhancement in 3D rendering pipelines, focusing on anti-aliasing and super-resolution.• Enhanced ML R&D infrastructure by developing a distributed training and evaluation framework; implemented a distributed data generation pipeline, reducing the time required for setting up the data generation by over 70%.• Improved training speed by 40% through custom Pytorch operations in CUDA.

  • AI Algorithm Engineer

    Aug 2022 - now

  • Deep Learning Software Engineer

    Jan 2021 - Aug 2022