- Built custom business analytics dashboards using Dash and Plotly, increased loading speed 10 times and enabling real-time data insights - Implemented a data synchronization pipeline between PostgreSQL and ClickHouse, reducing production database load by 17%. - Resolved client and internal team issues related to delivery zone mapping, order mispricing, and data reporting, ensuring smooth and uninterrupted operations - Led optimization of courier and warehouse pricing setup and cross-department coordination with accounting, resolving uncharged orders and invoices, and cutting negative cost variance -Developed repricing strategies and prediction algorithms for inventory metrics. Technologies: Clickhouse, SQL, supervised learning, cross-validation, hyperparameter tuning, model deployment, regression, gradient boosting

- Scaled export data reports to handle up to 1M rows and improved generation speed, streamlining clients’ accounting workflows - Integrated 7 new warehouses and couriers via REST API, SOAP, and GraphQL, enabling clients’ expansion across Europe, North America, and Australia - Enhanced system observability by implementing Metabase-based alerts and performance metrics, enabling early bug detection before client impact - Developed and deployed solutions for accurate parcel delivery address identification across multiple couriers, reducing invoice disputes by 12% and improving supplier cost predictability by 25% - Built interactive business intelligence dashboards in Metabase visualizing key performance indicators, enabling real-time monitoring of 25+ SLAs and significantly improving business transparency *Technologies*: Django, SQL, Celery, Redis, RabbitMQ, EDA, ETL, Matplotlib / Seaborn / Plotly

- Conducted R&D of state-of-the-art generative algorithms for optimal space planning - Implemented a graph-based spatial design module enabling automatic, optimized furniture layout generation - Developed unit and integration tests, increasing overall project test coverage by 20% - Defining business problem in terms of math model, design cost function. Implemented algorithms for architectural design solutions based on graph algorithms, evolutionary generative algorithms, machine learning. Technology: Python(Scipy, networkx, Tensorflow), Java (Spring), Unsupervised learning, Model training / evaluation

- Conducting data extraction, transformation via REST, SOAP, GraphQL, FTS APIs. - Developed an LSTM-based machine learning model to predict paper web breakages in advance, reducing stoppage frequency by 8% and increasing overall production output - Performed exploratory data analysis (EDA) on sensor data from oil well drills to identify and detect drill head slippage - Build analytical reports justifying solutions for stake holders and clients Technology: R(plotly, dplyr, ggplot), Python (Keras, Tensorflow, numpy, pandas, scipy), statistical inference, hypothesis testing

Prototyping of wireless ad-hoc modules on 2.4 and 5 GHz based on Raspberry PI, Orange PI PCs to construct mobile, agile mesh being able to transmit video signal.