Misha Usvyatsov

Misha Usvyatsov

General Chaos (Software Engineer)

Google

Biography

I am a Software Engineer at Google working on efficient on-device computing.

I finished PhD at ETH Zürich in the Photogrammetry and Remote Sensing Lab, supervised by Prof. Dr. Konrad Schindler.

Before moving to Switzerland I completed my masters in Moscow where I graduated with honors from Moscow Institute of Physics and Technology (MIPT) and from Skolkovo Institute of Science and Technology (Skoltech)

I started my expansion towards west with move to Kazan (where I spent 1 year at Innopolis) from Ekaterinburg (where I grew up and made my first bachelor degree in electrical engineering and graduated with honors from Ural Federal University).

Interests
  • Deep Learning
  • Machine Learning
  • Tensors
  • Computer Vision
Free time
  • SAMBO / Judo / BJJ
  • Paragliding
  • DIY (3D modelling, 3D printing, soldering, other)
  • Snowboarding, Swimming
  • Chess, Go
  • Coffee
Education

  • PhD in Deep Learning, 2022

    ETH Zürich, Zürich, Switzerland

  • MSc in Data Analysis, 2017

    Skolkovo Institute of Science and Technology (Skoltech), Moscow, Russia

  • MSc in System Programming, 2017

    Moscow Institute of Physics and Technology (MIPT), Moscow, Russia

  • BSc in Computer Science, 2015

    Innopolis University, Kazan, Russia

  • B.Eng. in Electrical Engineering, 2014

    Ural Federal University, Ekaterinburg, Russia

Experience

 
 
 
 
 
Software Engineer
Google
Nov 2022 – Present Zürich
Software Engineer at On-Device Performance Group
 
 
 
 
 
PhD Student / Research Assistant
ETH Zürich
Oct 2017 – Nov 2022 Zürich

Research areas: Deep Learning, Tensors, Computer Vision

Advisor: Prof. Konrad Schindler

 
 
 
 
 
Research Intern
Google
Jul 2021 – Oct 2021 Online
Research Intern at Mobile Vision Group
 
 
 
 
 
ML Intern
Apple
Apr 2020 – Aug 2020 Zürich
Research Intern at Special Projects Group
 
 
 
 
 
Software Engineering Intern
ProtonMail
Jun 2017 – Aug 2017 Online
  • Applied LSH for SPAM detection
  • Developed the system for import/export of emails for the ProtonMail system
 
 
 
 
 
Research Intern
Yandex
Jun 2016 – Jun 2017 Moscow

Worked with Prof. Andrey Ustyuzhanin in close collaboration with CERN.

  • Worked on muon tracks simulation with Generative Adversarial Networks
  • Developed efficient architecture (50x to 30% FLOPS speedup, depending on the problem)

Recent Publications

Quickly discover relevant content by filtering publications.
Mikhail Usvyatsov, Rafael Ballester-Rippoll, Lina Bashaeva, Konrad Schindler, Gonzalo Ferrer, Ivan Oseledets (2022). T4DT: Tensorizing Time for Learning Temporal 3D Visual Data. arXiv preprint arXiv:2208.01421.

Cite

Mikhail Usvyatsov, Rafael Ballester-Ripoll, Konrad Schindler (2022). tntorch: Tensor network learning with PyTorch. Journal of Machine Learning Research.

Cite

Anton Obukhov, Mikhail Usvyatsov, Christos Sakaridis, Konrad Schindler, Luc Van Gool (2022). TT-NF: Tensor Train Neural Fields. arXiv preprint arXiv:2209.15529.

Cite

Mikhail Usvyatsov, Anastasia Makarova, Rafael Ballester-Ripoll, Maxim Rakhuba, Andreas Krause, Konrad Schindler (2021). Cherry-Picking Gradients: Learning Low-Rank Embeddings of Visual Data via Differentiable Cross-Approximation. ICCV2021.

Cite PDF Code

Shengyu Huang, Zan Gojcic, Mikhail Usvyatsov, Andreas Wieser, Konrad Schindler (2021). PREDATOR: Registration of 3D Point Clouds with Low Overlap. CVPR2021.

Cite PDF Code

Shengyu Huang, Mikhail Usvyatsov, Konrad Schindler (2020). Indoor Scene Recognition in 3D. 2020 International Conference on Intelligent Robots and Systems (IROS).

Cite PDF Code

Mikhail Usvyatsov, Konrad Schindler (2019). Visual recognition in the wild by sampling deep similarity functions. 2019 International Conference on Robotics and Automation (ICRA).

Cite DOI PDF Code

Timo Hackel, Mikhail Usvyatsov, Silvano Galliani, Jan D Wegner, Konrad Schindler (2018). Inference, learning and attention mechanisms that exploit and preserve sparsity in convolutional networks. International Journal of Computer Vision.

Cite DOI PDF Code

Maxim Borisyak, Mikhail Usvyatsov, Michael Mulhearn, Chase Shimmin, Andrey Ustyuzhanin (2017). Muon trigger for mobile phones. Journal of Physics: Conference Series.

Cite DOI PDF Slides Code

Teaching

I taught the following courses:

  • Teaching assistant of Prof. Konrad Schindler.

    • Image Interpretation, Fall 2017 - 2021 Zürich, Switzerland

  • Teaching assistant of Prof. Stamatios Lefkimmiatis.

    • Signal and Image Processing, Spring 2017, Moscow, Russia

  • Course instructor.

    • Introduction to Deep Learning, May 2017, April 2021, Yerevan, Armenia
    • Introduction to Scientific Computing, Fall 2018 - 2021 Zürich, Switzerland

Recent Posts

Large-scale 3D modeling of a village Mulegns
The story of three weeks project as a part of Geodesic Project Course.
Jul 10, 2021 1 min read
Large-scale 3D modeling of a village Mulegns
Fixing Keychron K3 brown switch
How I fixed Keychron K3 brown switch.
Feb 25, 2021 3 min read
Fixing Keychron K3 brown switch

Contact