Architected scalable OCR workflows with Amazon Textract, extracting and cleaning tabular data from scanned PDFs for machine learning models, reducing manual processing time by 90%

Designed natural language, RAG-based interface enabling non-technical users to query complex healthcare datasets

Developed contrastive deep learning models with PyTorch to accurately predict metabolite presence in samples using mass spectrometry data

Analyzed and presented machine learning-based metabolomics research during weekly lab meetings

Automated timesheet data processing for 27 reference specialists, reducing report delivery time by 80%