Self-Supervised Learning for Medical Image Segmentation
Exploring contrastive pretraining strategies for scarce-label medical imaging tasks.
Active and completed collaborative research projects across computer vision, NLP, RAG, and large language model systems. Many are open for new collaborators - get in touch.
Exploring contrastive pretraining strategies for scarce-label medical imaging tasks.
Building evaluation benchmarks for retrieval tasks in low-resource language settings.
Designing and evaluating RAG pipelines that synthesize findings across research paper corpora.
Investigating parameter-efficient adaptation methods (LoRA, adapters) for scientific and technical text.
Building models that jointly parse figures, tables, and captions in academic papers for structured extraction.
To add a project, copy any <article class="pub"> block and set data-type to one or more space-separated values (computer-vision / nlp / rag / llm / ongoing). Multiple values let a project appear under all relevant filters.