RudolfV is a foundation model developed by pathologists for pathologists, designed to enhance the efficiency and performance of pathology foundation models by integrating pathologist expertise, semi-automated data curation, and a diverse dataset from over 15 laboratories, including 58 tissue types and 129 staining modalities. The model was trained on a dataset of 134,000 slides from 34,000 cases, curated by pathologists to maximize diversity across various parameters. The model outperformed existing state-of-the-art foundation models on multiple benchmarks, including tumor microenvironment profiling, biomarker evaluation, and reference case search, while demonstrating favorable robustness properties. The study highlights how domain-specific knowledge can improve the performance of pathology foundation models and enable novel applications. RudolfV was named in honor of Rudolf Virchow, a pioneer of modern pathology. The model was trained using a self-supervised learning approach, with data augmentation and clustering strategies to balance the representation of frequent and rare diseases. The model's performance was evaluated on various tasks, including tumor microenvironment characterization, immunohistochemistry biomarker scoring, and reference case search, where it achieved the best results on 10 out of 12 benchmarks and 28 out of 31 datasets. The model's robustness was also demonstrated across different staining and scanner types. The study shows that incorporating pathologist knowledge into foundation model development can lead to better performance and broader clinical applicability.RudolfV is a foundation model developed by pathologists for pathologists, designed to enhance the efficiency and performance of pathology foundation models by integrating pathologist expertise, semi-automated data curation, and a diverse dataset from over 15 laboratories, including 58 tissue types and 129 staining modalities. The model was trained on a dataset of 134,000 slides from 34,000 cases, curated by pathologists to maximize diversity across various parameters. The model outperformed existing state-of-the-art foundation models on multiple benchmarks, including tumor microenvironment profiling, biomarker evaluation, and reference case search, while demonstrating favorable robustness properties. The study highlights how domain-specific knowledge can improve the performance of pathology foundation models and enable novel applications. RudolfV was named in honor of Rudolf Virchow, a pioneer of modern pathology. The model was trained using a self-supervised learning approach, with data augmentation and clustering strategies to balance the representation of frequent and rare diseases. The model's performance was evaluated on various tasks, including tumor microenvironment characterization, immunohistochemistry biomarker scoring, and reference case search, where it achieved the best results on 10 out of 12 benchmarks and 28 out of 31 datasets. The model's robustness was also demonstrated across different staining and scanner types. The study shows that incorporating pathologist knowledge into foundation model development can lead to better performance and broader clinical applicability.