Viktor Sukhanov

Redesigned the bank integration core feature, reducing new bank integration time from 2 days to 2 hours. Developed core multi-currency payment feature, expanding into other regions, driving 20-30% profit increase. Developed a refactoring and stabilization roadmaps. Optimized development workflows, reducing both overtimes and critical release bugs by 95%*. *Based on statistics of the amount of night calls from CTO and developers feedback. Managed 2 teams 8 people each: backend, frontend, QA engineers.

Built a MLOps platform called DSW that helps ML engineers start virtual environments quickly with any necessary services out of the box, accelerating ML experiments by 15% across the company. *15% is based on feedback from project team leads. Developed multiple python tools to automate repetitive ML tasks, accelerating engineers workflow. Optimized data processing pipelines, improving efficiency of ML development across the company.

Built a product price forecasting model based on boosting algorithms and fbprophet seasonal coefficients, cutting prediction errors by 3-6% and enhancing operational planning Implemented a data preprocessing pipeline using Azure, Hadoop, Spark Deployed automated price forecasting model to production on Azure, generating 100-150M RUB in additional annual profit

Improved ML algorithms for petroleum product price forecasting by scraping more data, reducing the number of garbage features and stacking the predictions of different types of models, reducing prediction error by 4-5%