Sergei Tsereshko

I’m a Data Scientist who builds ML systems that help businesses make better decisions under uncertainty. For more than four years, I’ve worked in consumer lending, where model quality only matters if it leads to the right balance of growth, risk, speed, and cost.

The problems I enjoy rarely begin with an algorithm or end with a model. They start with defining the right objective and understanding the constraints, and they continue long after deployment. I like owning that whole path, especially the trade-offs that appear when a model becomes part of a real decision process.

In practice, I’ve seen a better scoring model add 7–8 Gini points, a carefully designed decision rule raise approval by 4 percentage points without adding risk, and a well-structured experiment turn a scoring change into a 1.5% profit lift. I’m equally interested in the quieter wins around the model: faster decisions, reliable monitoring, and smarter use of external data.

Most of this work happens in Python and SQL, with PySpark on large-scale data and Airflow in production. Credit risk is the domain I know best; the underlying problems—predictive modeling, experimentation, monitoring, and decision optimization—translate far beyond lending.

I’m open to Data Scientist opportunities where ML is connected to a real decision, the outcome is measurable, and I can stay involved from problem framing through production.