Senior Data Scientist at BCG X team with primary focus on predictive maintenance, operations analysis, and model production deployment. Notable cases include the following:• Developed predictive maintenance algorithm for food manufacturer that detects upcoming failure multiple hours in advance and notify factory personal about recommended preventive actions. • Designed and performed a plan for production migration from legacy deployment to AWS with full CI/CD automation. • Implemented algorithm for cost optimization of individual car parts for automotive industry client.

• Created, implemented and deployed a real-time pipeline leak detection system, that have monitored 950+ oil, water, and gas pipelines. It allowed to minimize the time of reaction on leak events to less than an hour for pinhole leaks, generated $10MM+ in annual value, reduced environmental footprint, and successfully replaced 3rd party software.• Developed a streaming pipeline that collects real-time video from IP cameras, process it using computer vision (YOLO, MaskRCNN) algorithms, and submits analytics for further decision making about abnormal activity at company's property.• Trained and deployed convolutional neural net (CNN) that classifies rod pump dynamometer cards in real-time.• Developed a geometrical algorithm that finds optimal well package placing inside a giving lease. Algorithm optimally place wells to maximize spacing and horizontal length up to required limit, and follows lease boundary requirements.• Implemented an algorithm for automatic segmentation of oil production curves with subsequent fit of each segment with either hyperbolic, exponential, or linear curve. The last segment was extrapolated to predict expected future oil production.

Developed a new, chemical knowledge-based algorithm for structure generation of amorphous materials. The generated samples reproduce structural features that would take at least 100,000 years for conventional methodologies to recover and provide key insight into long-standing problems of the first sharp diffraction peak, photodarkening, Boson peak, and the topological midgap electronic states. Methodology was implemented in Python and Matlab with utilization of many algorithms including Nelder-Mead, Dijkstra's algorithm, Delaunay triangulation, Voronoi tessellation.

Was a teacher assistant in a laboratory where I led physical chemistry experiments for undergraduate students from Honors College. Among my responsibilities were brief introductory lecture, control of experiment flow, and students grading.

Worked on the acoustic methods of high-energy neutrino detection. Developed an algorithm that use cross-correlation technique to calculate shift between audio signals, and triangulation to estimate the direction where signal came from w.r.t. the position of acoustic sensors. Signals that passed through the Earth considered as reliable and subject for further analysis, while those that came from the atmosphere were ignored. Used C programming language and ROOT package libraries to analyze more than 100,000 acoustic tracks. Successfully identified one high-energy neutrino events in the data set.