Dmitry Turchenkov 2017 — present Sberbank Senior Scala developer • Machine Learning Framework Implementation - a centralized model lifecycle management system integrated into the Hadoop ecosystem with support for Spark, Hadoop Streaming, Python, SAS. (Scala, Akka Http, Kerberos, Apache Oozie) • Implementing in-memory computation interface for native deep learning Python models. (GridGain, Apache Ignite, Docker, Python) • Data-source integration. (Kafka, Flume, Spark Streaming) • Kerberos web-authentication. (SSO, SPNEGO, Kerberos) 2016 — 2017 Cognitive Technologies Mathematician developer • Implementation of a path-planning algorithm combined with holonomic bicycle model constraints based on Reeds-Shepp curves with obstacles extension for advanced driver assistance system (ADAS) “C-Pilot”. (C++, CUDA, OpenCV, cuBLAS) • Multisensory integration module (GPS, LIDAR/LEDDAR) based on Extended Kalman filter; Drivability map estimation algorithm with stereo vision observation and Bayes filter modification. (C++, CUDA) • Traffic sign recognition system based on convolutional neural networks, implemented in “C-Pilot” ADAS. (C++, Anaconda/Python, Caffe, NVIDIA-DIGITS) May 2015 — October 2015 Megaputer Intelligence Mathematician developer • Implementation big-data algorithms in “Polyanalyst” on GPU architecture based on NVIDIA CUDA. (T-test, ANOVA, Decision tree, SVM, Multiple linear regression) • GPU-Implementation self-organized neural network combined with genetic algorithms for cluster analysis. (RPROP, Quickprop) • Spiking neural networks based on LIF-model and spike-timing dependent plasticity (STDP) were implemented in library. (C++, CUDA) 2013 — 2015 Institute of Mathematical Problems of Biology of Russian Academy of Sciences Ph.D. Student • Combined approach implementation of hybrid molecular-Brownian dynamics and modern DFT methods of quantum chemistry to computer simulation of ionic channel conductivity process in a lipid bilayer. Proposed approach implemented in PCS (Patch Clamp Simulation) software with GPU acceleration support, based on NVIDIA CUDA, and interactive 3D-visualisation. (DFT, Molecular Dynamic, C++, CUDA) 2013 — 2014 Molecular Simulations Group, Lomonosov Moscow State University Research fellow developer • Develop and apply modern computer simulation methods to study biological and nano-systems at the molecular level. (Molecular dynamics, quantum chemistry, QM/MM, Voreen, coarse-grain simulations) 2013 — 2014 НИЯУ МИФИ, кафедра высшей математики Lecturer • Lectures, training courses and special analytical studies. (Linear algebra, Mathematical analysis, Analytical geometry)