Karol Duris

I’m a Data Scientist who’s passionate about uncovering insights from data and translating them into real-world solutions. My work blends supervised and unsupervised learning with practical pipeline engineering so models drive measurable outcomes.

I’ve built predictive scoring and automation in recruitment analytics—developing models to forecast candidate success probability and improving sourcing effectiveness from 22% to 46%. I’ve also delivered customer behavior modeling for driving ancillary product attachment, contributing to a more than 2 increase in attachment rate.

Across domains, I’ve led predictive and anomaly-focused modeling: from early detection of improperly wound rolls using high-frequency sensor time-series to credit scoring and credit-limit forecasting with transactional and behavioral data. I focus on robust feature engineering, iterative modeling, and dependable experimentation to support decision-making.

I also bring hands-on GenAI experience, working with Vertex AI on GenAI LLMs and building LLM-powered assistants and agentic workflows using LangChain and OpenAI APIs. I enjoy pairing strong analytics fundamentals with modern AI tooling—so teams can ship smarter, faster.