Tamara Atanasoska works on ML explainability, interpretability, and fairness as an Open Source Software Engineer at :probabl., a company at the heart of the open source scikit-learn ecosystem. She is a maintainer of Fairlearn, the open-source library for assessing and improving the fairness of machine learning systems, and a contributor to scikit-learn and skops, placing her among the people shaping the tools that thousands of practitioners rely on every day.

Tamara’s work covers the practical side of responsible AI: group fairness metrics and mitigation methods, the trade-offs between false positives, false negatives, and demographic parity, interpretability tools such as partial dependence, and secure model serialization. She approaches fairness as a sociotechnical problem, not just a metric, emphasizing human-in-the-loop systems and the organizational questions of who should decide fairness trade-offs.

She holds a background in both Computer Science Software Engineering and Computational Linguistics (NLP), a combination that informs how she thinks about models, language technologies, and the people affected by them. Tamara also writes about language and technology on her Substack, Holophrase.