PRICOPE, T.-V. HardML: A Benchmark for Evaluating Data Science and Machine Learning Knowledge and Reasoning in AI. Studia UBB Informatica 2025, 69, 59-76.