Maximilian Werk developed multilingual document retrieval benchmarks and enhanced embedding model support across several repositories. In embeddings-benchmark/mteb, he created the JinaVDR benchmark dataset, adding diverse tasks and configurations for multilingual and layout-rich document types, leveraging Python and data engineering skills to ensure robust dataset integration. For qdrant/landing_page, he updated documentation and onboarding materials to reflect new Jina v4 embedding model capabilities, improving developer experience and integration clarity. In ggml-org/llama.cpp, Maximilian implemented EuroBERT model support within the C++ backend, enabling GGUF conversion and model serialization. His work demonstrated depth in C++ development, machine learning, and technical documentation.