It does mean a thing

Some notes on software and beyond

Sergey Lisitsyn


  • 2012 – 2014, MSc, Samara State Aerospace University, Samara, Russia.
    Thesis: Inductive transfer of classification models.
    Investigated the problem of inductive transfer via non-linear dimensionality reduction. Formulated kernel-based variation of the local tangent space alignment dimension reduction algorithm. Developed a fast parallel implementation of the proposed method [2].
  • 2008 – 2012, BSc, Samara State Aerospace University, Samara, Russia.
    Thesis: Traffic sign recognition using support vector machines.
    Composed a traffic sign recognition model based on the multiclass support vector machine classifier, the histogram of oriented gradients features and the homogeneous kernel mapping procedure [1]. The model has been successfully (top performance for non-neural methods back then) applied to the German Traffic Sign Recognition Benchmark (GTSRB) dataset.



  • 2014 – present, Software engineer, Yandex, Moscow, Russia.
    Development of internet-scale machine learning infrastructure for advertisement. Extensive experience of data analysis, high-performance systems and practical machine learning.
  • 2012 – 2014, Software engineer, Smart Solutions, Samara, Russia.
    Initial design and further development of a scheduling system based on high-throughput messaging. Development of a web-based railroad scheduling application from the scratch. Comprehensive experience of concurrent systems design, optimization and basics of team management.
  • 2011 – 2012, Software engineer, NetCracker, Samara, Russia.
    Defects resolution, documentation improvements and various development tasks related to an operations support system that is made on top of the Java EE stack. Decent experience of large-scale systems engineering support.



Open Source

  • One of the core developers of the Shogun machine learning toolbox.
    Google Summer of Code mentor for projects:
    • Implement algorithms for Blind Source Separation (BSS) (done in 2013 by Kevin Hughes),
    • Essential Deep Learning Modules (done in 2014 by Khaled Nasr)
  • Author of Tapkee, a C++ library for dimensionality reduction.
  • Author of a few other C++ libraries: formatting, aer, algorithme, stichwort.


  • Machine learning: kernel and linear methods, dimensionality reduction, feed-forward neural networks
  • Software engineering: C++, Java and Python; distributed systems, algorithms and high-performance computing
  • Some Kaggle experience: Avazu Click-Through Rate Prediction (top 10% of 1604)