Yura Kuratov

deep learning researcher / memory-augmented models

I work on memory-augmented neural architectures to address limitations of transformer-based models in long-context processing, sequential computation, and test-time adaptation, while also making memory representations more compact. My research also covers external memory for LLMs, including retrieval-augmented generation and knowledge graphs. My broader research interests include genomic language models, where long sequences, biological structure, and complex data call for new modeling approaches.